Systems Biology, Bioinformatics, and Biomarkers in Neuropsychiatry

REVIEW ARTICLEpublished 24 December 2012doi 103389fnins201200187

Systems biology bioinformatics and biomarkers inneuropsychiatryAli Alawieh1 Fadi A Zaraket 2 Jian-Liang Li 3 Stefania Mondello4 Amaly Nokkari 1 Mahdi Razafsha5 BilalFadlallah6 Rose-Mary Boustany 17 and Firas H Kobeissy 151 Department of Biochemistry College of Medicine American University of Beirut Beirut Lebanon2 Department of Electrical and Computer Engineering Faculty of Engineering and Architecture American University of Beirut Beirut Lebanon3 Sanford-Burnham Medical Research Institute Orlando FL USA4 Department of Neuroscience University of Messina Messina Italy5 Division of Addiction Medicine Department of Psychiatry Center for Neuroproteomics and Biomarkers Research University of Florida Gainesville FL USA6 Department of Electrical and Computer Engineering University of Florida Gainesville FL USA7 Department of Pediatrics American University of Beirut Medical Center Beirut Lebanon

Edited byXiaogang Wu IndianaUniversity-Purdue UniversityIndianapolis USA

Reviewed byScott H Harrison North Carolina AampTState University USASudipto Saha Bose Institute IndiaHussam Jourdi Florida StateUniversity USADan Xia Harvard Medical School USA

CorrespondenceFiras H Kobeissy Division ofAddiction Medicine Department ofPsychiatry Center forNeuroproteomics and BiomarkersResearch University of FloridaL4-100F PO Box 100256 GainesvilleFL 32610 USAe-mail firaskogmailcom

Although neuropsychiatric (NP) disorders are among the top causes of disability worldwidewith enormous financial costs they can still be viewed as part of the most complex disor-ders that are of unknown etiology and incomprehensible pathophysiology The complexityof NP disorders arises from their etiologic heterogeneity and the concurrent influence ofenvironmental and genetic factors In addition the absence of rigid boundaries betweenthe normal and diseased state the remarkable overlap of symptoms among conditionsthe high inter-individual and inter-population variations and the absence of discriminativemolecular andor imaging biomarkers for these diseases makes difficult an accurate diag-nosis Along with the complexity of NP disorders the practice of psychiatry suffers from aldquotop-downrdquo method that relied on symptom checklists Although checklist diagnoses costless in terms of time and money they are less accurate than a comprehensive assess-ment Thus reliable and objective diagnostic tools such as biomarkers are needed thatcan detect and discriminate among NP disorders The real promise in understanding thepathophysiology of NP disorders lies in bringing back psychiatry to its biological basis ina systemic approach which is needed given the NP disordersrsquo complexity to understandtheir normal functioning and response to perturbationThis approach is implemented in thesystems biology discipline that enables the discovery of disease-specific NP biomarkers fordiagnosis and therapeutics Systems biology involves the use of sophisticated computersoftware ldquoomicsrdquo-based discovery tools and advanced performance computational tech-niques in order to understand the behavior of biological systems and identify diagnosticand prognostic biomarkers specific for NP disorders together with new targets of thera-peutics In this review we try to shed light on the need of systems biology bioinformaticsand biomarkers in neuropsychiatry and illustrate how the knowledge gained through thesemethodologies can be translated into clinical use providing clinicians with improved abilityto diagnose manage and treat NP patients

Keywords systems biology biomarkers bioinformatics psychiatry data mining proteomics autism omics

INTRODUCTIONNeuropsychiatric (NP) disorders like Schizophrenia (SZ) MajorDepression Disorder (MDD) Bipolar Disorder (BPD) and Obses-sive Compulsive Disorder (OCD) are among the top causes ofdisability worldwide (Lopez and Murray 1998) The WHO reportpublished in 1998 concerning the top causes of disability expectedfor the year 2020 had four out of 10 diseases being NP diseaseswith depression ranking first and BPD ranking seventh (Lopezand Murray 1998) The 2004 update of the report states thatNP disorders account for one third of the top causes of disabilityworldwide with unipolar depression disorder still ranking first forboth sexes (WHO 2009) A recent report commissioned by theEuropean Brain Council (Gustavsson et al 2011) concluded that

brain disorders cost Europe almost C800 billion (US$1 trillion) ayear ndash more than cardiovascular disease diabetes and cancer puttogether No directly comparable reports exist elsewhere in theworld but several studies looking at the costs of individual con-ditions such as BPD attention deficit hyperactivity disorder andSZ in both the United States and Europe have shown that health-care costs per person are similar in both regions (Smith 2011)Simultaneously there has been an increased appreciation of thecomplexity of NP diseases and the inadequacy of the availablediagnostic prognostic and therapeutic approaches that dependmainly on clinical diagnosis and lacks disease-specific molecularbiomarkers (Linden 2012) The complexity of NP disorders arisesfrom the high level of etiologic heterogeneity and involvement of

wwwfrontiersinorg December 2012 | Volume 6 | Article 187 | 1

Alawieh et al Bioinformatics and systems biology in neuropsychiatry

several multifactorial environmental and genetic factors (Schorket al 2007 Cacabelos et al 2011 Gormanns et al 2011 Mitchell2011 Ripke et al 2011) Unlike classical Mendelian inherited dis-easesa single gene mutation cannot explain the overt phenotype ofNP disorders Clinical appearances generally result from the con-current influence of different genes along with several epigeneticmechanisms (Cowan et al 2002 Behan et al 2009 Cacabeloset al 2011)

Many other limitations circumscribe the current study of NPdisorders including the inaccessibility to the brain tissue (Levinet al 2010 Villoslada and Baranzini 2012) the overlap of symp-toms among conditions that can result in diagnostic ambiguityand the inability to identify specific diagnostic biomarkers usingcurrent imaging techniques like functional Magnetic ResonanceImaging (fMRI) Positron Emission Tomography (PET-scan) Sin-gle Photon Emission Computed Tomography (SPECT) and others(refer to the review by Linden 2012) Moreover current avail-able diagnostic systems like DSM-IV and ICP-10 rely principallyon cliniciansrsquo assessment even if different disorders ndash treated dif-ferently ndash have similar manifestations (Linden 2012 Tretter andGebicke-Haerter 2012) Molecular biology techniques may enablethese diagnostic strategies to be more accurate and objective andallow for disease classifications better suited for more targetedpersonalized and effective therapy

With these aims in mind the previous molecular biologyapproach that investigates the effect of a single gene or proteinappears insufficient to establish a complete understanding of theNP disorders (Kitano2002 Zhang et al 2010) Such a reductionistapproach fails to provide an insight into the structure and dynam-ics of biological systems Biological systems are characterized byhierarchy (having several layers of organization with interactionsamong and between layers) emergence (having properties thatare visible at the system level and unexpected from the underly-ing components alone) and robustness (having the capacity ofthe system or network to operate normally under a wide range ofconditions) Meanwhile the reductionist approach lags behind theability to detect emergent properties and combinatorial effects thatcharacterize biological systems It also falls short of understand-ing compensatory and synergistic responses of the system towardperturbations (Lucas et al 2011 Munk et al 2011 Saetzler et al2011 Westerhoff 2011) Systems biology would address the mul-tifactorial aspect of NP diseases to better chart the functionalityof complex biological systems for insights into the etiology andpathophysiology

In this review we will be discussing the potentials of systemsbiology bioinformatics and biomarker research in the area of NPdisorders Furthermore we will be outlining how the knowledgegained through these methodologies can be translated into clin-ical use providing clinicians with improved ability to diagnosemanage and treat NP patients

THE APPROACH OF SYSTEMS BIOLOGYThe revolution of systems biology in the twentieth century rep-resented a ldquoparadigm shiftrdquo of molecular biology from a reduc-tionist approach to a holistic approach (Westerhoff and Palsson2004) The new approach tries to analyze the relationships withina biological system to find how different components of that

particular system interact (Kitano 2002 Hood and Perlmutter2004 Fang and Casadevall 2011 Lucas et al 2011 Westerhoff2011 Karsenti 2012) This accelerating ldquoshiftrdquo was grounded ontwo lines of thought whose integration is believed to have trans-formed molecular biology into the more sophisticated disciplineof systems biology (Westerhoff and Palsson 2004) These are thesequencing of the human genome and the great advances in thehigh-throughput (HT) discovery tools that allowed collection andanalysis of large data sets of genomics proteomics and others(Ideker et al 2001 Kitano 2002 Food and Drug Administra-tion 2011 Ori et al 2011) It is believed that systems biology willhelp understand and simplify the complexity of biological systemsand the available datasets obtained by HT techniques Such com-plexity is impeding the development of new diagnostic tools andtherapeutics for several complex diseases especially in the areas ofneuropsychiatry (Hood and Perlmutter 2004 Robeva 2010 Fangand Casadevall 2011 Westerhoff 2011)

Systems biology works by combining mathematical modelswith experimental molecular information from in silico in vivoand in vitro studies with HT data sets including genomicsproteomics metabolomics and transcriptomics (Robeva 2010Zhang et al 2010 Westerhoff 2011) For this purpose the integra-tive use of computational tools bioinformatics and engineeringsystems analysis represents the working tools in systems biologyTherefore systems biology requires a strong computational infra-structure and simulation software tools (Kitano 2002 Hood andPerlmutter 2004) that are able to handle huge databases and deter-mine dependencies that can be correlated with biological functions(Jamshidi and Palsson 2006) In addition HT genomics pro-teomics and metabolomics infrastructure are needed to achieverobustness and reproducibility (Kitano 2002 Hood and Perl-mutter 2004 Westerhoff and Palsson 2004 Jamshidi and Pals-son 2006) Finally enough experimental data should be gath-ered to provide raw material for analysis and to validate presentresults generated from multidisciplinary fields of mathematicsengineering bioinformatics and medicine (Robeva 2010)

The holistic analysis of systems biology aims to solve the previ-ously mentioned limitations of the reductionist approach Theseinclude (1) the inadequacy to investigate the hierarchy robust-ness and emergence characteristics of a biological system (2)the incompetence in understanding and reconstituting systemrsquosdynamics and (3) the failure to decipher the complexity of themassive data output from HT techniques (Kitano 2002 Zhanget al 2010 Lucas et al 2011) Therefore systems biology pro-vides a mean to understand the normal functioning of the systemand to predict the systemsrsquo response to perturbations (Idekeret al 2001 Kobeissy et al 2008) With this approach systemsbiology increases diagnostic prognostic and disease-monitoringpotentials for clinical applications (Hood and Perlmutter 2004)

The holistic approach of systems biology can be either a top-bottom approach starting from ldquoomicsrdquo datasets and drawinginferences related to the flow of information within biological net-works or bottom-top approach starting from experimental mole-cular data to draw models of these networks (Jamshidi and Palsson2006 Fang and Casadevall 2011 Lucas et al 2011 Wester-hoff 2011) The networks and sub-networks that systems biologyaims to study represent modules in the biological systems These

Frontiers in Neuroscience | Systems Biology December 2012 | Volume 6 | Article 187 | 2


biological networks are characterized by non-linear interactionsbetween components providing the floor for organization andstructure (Ideker et al 2001) Accordingly most biological net-works adopt a scale-free network characterized by a ldquopower-lawrdquodistribution where the majority of nodes (network components)have few links and only few nodes called ldquohubsrdquo have a highnumber of links (Zhu et al 2007 Saetzler et al 2011) Such orga-nization has been associated with biological networks includingthe large and diverse proteinndashprotein interaction network Thisnetwork has most of its regulation occurring at the level of theldquohubsrdquo which are master protein regulators It is also characterizedby the presence of buffers that help mitigate the effect of any poten-tial noise in the system by preventing it from activating biologicalprocesses By studying and modeling these networks researcherscan identify key nodal proteins and their interactions along withthe major pathways involved These components and pathwayswould be the major players in the pathophysiology of a disease ordisturbance and are the first suggested targets of therapy (Hoodand Perlmutter 2004 Zhu et al 2007)

THE APPLICATION OF SYSTEMS BIOLOGY INNEUROSCIENCE AND PSYCHIATRYThe major quest in the field of NP disorders is to reduce behav-ioral and mental phenomena into biochemical properties that canbe tested and examined objectively through molecular biologytechniques (Tretter and Gebicke-Haerter 2012) This is carriedout by the application of systems biology to NP disorders andneuro-systems starting by the integration of HT datasets fromthe different levels of the system (Zhang et al 2010) Disciplinesinclude genomics transcriptomics epigenomics and proteomicsall of which can gather information on neurosystem structure anddynamics in normal and disordered states at different echelonsstarting with organelle synaptic and neuronal levels and pro-ceeding to brain circuitry and sub-networks up to the entire brainlevel as illustrated in Figure 1

The application of theseldquoomicsrdquo approaches can be achieved byeither the HT techniques to constitute new datasets under desiredcircumstances or through literature or data mining using high per-formance algorithms discussed later Systems biology incorporatesbrain imaging techniques in vitro and in vivo studies of neuronalcell performance in different conditions and in silico modelingtechniques These in silico modeling techniques can demonstratecertain pathological or normal states and help draw inferencesabout the dynamicity of the system and the perturbations causedby drugs and interventions All these data sources help to drawaccurate mathematical models of the interactions of the perturbedsystem from which an insight can be taken into the physiology ofthe system and pathophysiology of diseases (Robeva 2010 Zhanget al 2010 Westerhoff 2011)

SYSTEMS BIOLOGY APPLICATION IN SCHIZOPHRENIASystems Biology has been employed to investigate the patho-physiology of complex diseases like SZ Several hypotheses wereproposed to explain the etiology and pathophysiology of SZ Thesehypotheses include among others (1) the neurodevelopmentalhypothesis that explains SZ in terms of abnormalities in perinataldevelopment (Arnold et al 2005 Hayashi-Takagi and Sawa 2010

FIGURE 1 | Components of neuro-systems biology The different sourcesof information for the application of systems biology to NP diseases thecomponents that are a common target of HT-discovery tools have beengrouped together

Altamura et al 2012 Khandaker et al 2012 Miller et al 2012)(2) the neurodegenerative hypothesis that links SZ to excitotoxicitycaused by hyperglutaminergic signaling (Malaspina 2006 Perez-Neri et al 2006 Kitabayashi et al 2007 Archer 2010) (3) theclassical dopamine hypothesis that associates SZ with dopaminesignaling dysfunction including hypo-dopaminergic signaling inthe prefrontal cortex and hyper-dopaminergic signaling in themesolimbic system (Snyder 1976 Sayed and Garrison 1983 See-man 1987 Baumeister and Francis 2002) (4) the glutamatehypothesis that associates SZ with hyperglutaminergic signaling(Coyle 2006 Stahl 2007 Egerton and Stone 2012 Moghad-dam and Javitt 2012) and (5) the revised dopamine hypothe-sis that includes both glutaminergic signaling hyperactivity anddopaminergic signaling dysfunction Systems biology has beenincorporated to assess and provide evidence for current or newhypotheses of SZ disease (King et al 1981 da Silva Alves et al2008 Howes and Kapur 2009) One example of this applicationinvolves the study of neurotransmitter (NT) dysfunction includ-ing dopaminergic GABA-ergic and glutaminergic transmission(discussed later)



As supported by imaging techniques and postmortem analysisSZ is characterized by loss of inhibitory interneurons in severalbrain regions involving GABA transmission disrupting both theglutaminergic and dopaminergic transmission (Freedman 2003Martins-de-Souza 2010) The use of fMRI supports the idea ofloss of inhibitory interneurons in the hippocampus and dorso-lateral prefrontal cortex as indicated by hyperactivity in theseareas (Freedman 2003) Consequently both hyperglutaminer-gic transmission and dopaminergic transmission dysfunctions aremajor hypotheses to explain the pathophysiology of SZ More-over in an in vitro study by Martins-de-Souza et al the treatmentof astrocytes with MK-801 an N -Methyl-d-aspartate (NMDA)antagonist revealed proteomic changes similar to those observedin SZ human brain tissue The use of clozapine an antipsychoticreversed these changes (Martins-de-Souza et al 2011)

Several meta-analyses of Copy-Number Variants (CNV) andSingle Nucleotide Polymorphism (SNP) studied susceptibility locifor SZ The studies performed by Allen et al (2008) Crespi et al(2010) involving the SZGene database revealed the involvementof deletions affecting the loci of Catechol-O-Methyltransferase(COMT) gene responsible for dopamine metabolism dopaminereceptors DRD1 and DRD2 and genes related to GABA andGlutamate signaling Similar associations of these genes weredetermined by Major Depressive Disorder Working Group of thePsychiatric GWAS Consortium (2012) using BEAGLE 304 a soft-ware package for analysis of large-scale genetic data to impute 12million autosomal SNP On the other hand Abdolmaleky et alstudied altered methylation patterns of postmortem brain samplesfrom frontal cortex using univariate bivariate and multivariatestatistical tests with large sample approximations to assess differ-ences between cases SZ or BPD and controls Results indicatethat the hypomethylation of the COMT gene was associated withmRNA overexpression of the gene and decreased DRD1 expressionin both SZ and BPDs (Abdolmaleky et al 2006) Similar statisti-cal analysis was performed by Nohesara et al who used saliva acomparable non-invasive source for DNA study to blood to eval-uate the methylation patterns in human SZ and BPD subjects (formore information about the use of saliva as a DNA source checkAbraham et al 2012 Simons et al 2012) Results showed thatthe same gene COMT was also hypomethylated (Nohesara et al2011)

Moreover in the study done by Kirov et al (2008) CNVs in SZwere analyzed by comparative genome hybridization using theCGHPRO software for the analysis and visualization of arrayCGH data Results showed that deletion at the loci of NRXN1a neuronal cell adhesion molecule that modulates the recruitmentof NMDA receptor (Lett et al 2011) and de novo duplication inthe amyloid precursor-binding protein A2 gene (APBA2) are asso-ciated with SZ The association of CNVs involving the NRXN1 wasreported in another study done by Kirov et al (2012) In this newstudy the Affymetrix Genotyping Console 40 software was usedto integrate and visualize data related to CNVs implicated in SZAlong with the locus of NRXN1 the study implicated seven otherloci mainly related to post-synaptic density (PSD) proteins andNMDAR signaling These data were subject to a systemic biologyapproach that related the results of CNV studies to those of pro-teomic analysis that showed PSD-proteome enrichment (Kirov

et al 2012) In addition another study by Lett and colleaguessupports this finding by associating a SNP at NRXN1 locus withSZ This study involved the statistical analysis of gene associationusing UNPHASED 31 software for genetic association analysis(Lett et al 2011) Furthermore a study by Le-Niculescu et alinvolved the integration of pharmaco-genomic mouse model withhuman genetic linkage data and human postmortem brain data Inthis study genomic and pathway analysis involved the use of Gene-Spring software for hierarchical clusteringNetAffx Gene OntologyMining Tool for categorizing genes into functional categories andIngenuity Pathway Analysis to analyze direct interactions amongtop candidate genes Results showed the involvement of pathwaysrelated to GABA transmission glutamate transmission synapticsignaling myelination and lipid metabolism (Le-Niculescu et al2007)

In addition to the above-mentioned genomic analyses severalproteomic studies of postmortem brain tissues from several corti-cal areas of SZ patients also supported the implication of dopamin-ergic glutaminergic and GABA-ergic transmission dysfunction inthe pathophysiology of SZ (Pennington et al 2008 Behan et al2009 Nesvaderani et al 2009 Martins-de-Souza 2010) The studydone by Pennington et al (2008) included a proteomic analysis ofthe proteome of layer 2 of the insular cortex using DeCyder 50 forstatistical differential analysis and XTandem for protein identifi-cation Results involved 57 differentially expressed spots of which17 of which being related to cellndashcell communication and signaltransduction Glial fibrillary acidic protein (GFAP) one of the pro-teins related to glutaminergic NMDA signaling has been shown tobe differentially expressed in several regions of postmortem brainof SZ patients These regions includedorsolateral prefrontal cortex(Martins-de-Souza 2010) the Wernickersquos area (Martins-de-Souza2010) the anterior temporal lobe (Martins-de-Souza 2010) thehippocampus (Focking et al 2011) and the anterior cingulatecortex (Clark et al 2006) Another protein (PRDX6) a phos-pholipase 2 associated with reduced dopaminergic transmissionhas been implicated in SZ by linkage study of Hwu et al (2003)at the locus 1q251 and in proteomic studies involving severalbrain regions (Nesvaderani et al 2009 Martins-de-Souza 2010)These ldquoomicsrdquo studies are consistent with the previous hypothesissupporting the involvement of NMDR receptor signaling dysfunc-tion GABA-ergic hypofunction and dopamine dysfunction in thepathophysiology of SZ

The relation between the loss of GABA inhibition hyperglut-aminergic signaling and dopamine transmission dysfunction hasbeen emphasized by studies done by Martins-de-Souza (2010)evaluating calcium homeostasis-related proteins in SZ The studyinvolved network analysis using the STRING software to drawproteinndashprotein interactions implicated in the pathogenesis ofSZ This analysis shows that in SZ the disruption of proteinsrelated to calcium balance including glutamate may increase Ca+2-dependent phospholipase A2 (PLA2) activity and account forthe accelerated phospholipid turnover and reduced dopaminer-gic activity seen in the SZ frontal lobe (Martins-de-Souza 2010)In vivo studies on animal models further support these hypothesesinvolving the signaling systems in SZ utilizing NMDAR antag-onists like PCP on both mice (Enomoto et al 2007) and rats(Pollard et al 2012)



Studies of GLAST-knockout mice (GLAST is a protein involvedin glutamate clearance) supports the hypothesis that glutamatehyper-transmission contributes to the pathophysiology of SZ(Karlsson et al 2008 2009) One study by Karlsson et al (2009)on mice reveals that GLAST-knockout produce phenotypic abnor-malities related to the negative and cognitive symptoms similar toSZ patients Moreover the same group showed similar results ina previous study on GLAST-knockout mice and supported theantipsychotic potential of mGlu23 agonists that decrease Gluta-mate release from neurons (Karlsson et al 2008) Beyond thisintegration systems biology further expands to the predictivepart where several computational models have been designedto mimic circuit alterations in SZ and provide a platform tounderstand the pathophysiology of the disease and the effectsof drugs and therapeutics as will be discussed in the in sil-ico modeling section (Vierling-Claassen et al 2008 Spencer2009 Rotaru et al 2011 Volman et al 2011 Komek et al2012)

However the application of systems biology in SZ is not limitedto the study of synaptic transmission pathways associated with thedisease other pathways have been implicated in the pathophysiol-ogy of SZ involving neurodevelopment and synaptic plasticity cellcytoskeleton abnormalities signal transduction pathways cellu-lar metabolism and oxidative stress patterns of myelination andinflammation and cytokine production We will briefly demon-strate the results of the study of association between SZ andneurodevelopment although a detailed review of all the models isbeyond the scope of this review

The association between neurodevelopmental changes and SZhas been reported in several genomic transcriptomic and pro-teomic studies (Rapoport et al 2005) The GWA study launchedby the Psychiatric Genome-Wide Association Study (GWAS) Con-sortium (2011) mentioned before implicated MIR137 as a newloci associated with SZ at the genome-wide-significance (GWS)level miRNA 173 is a known regulator of neuronal development(neurogenesis) Transcription factor 4 (TCF4) CACNA1C andCALN1 (calneuron 1) are miRNA 173 predicted targets and arealso associated with SZ (Ripke et al 2011) The study of Green-wood et al (2011) that analyzed several SNPs association to neuro-physiological and neurocognitive endophenotypes of SZ revealedthe involvement of several genes involved in axonal guidance(AKT-1 BDNF) and neurodevelopment (ERB4 NRG-1 Green-wood et al 2011) Implication of genes associated with SZ inneurodevelopmental changes was demonstrated by the transla-tional convergent functional genomics (CFG) study conductedby Ayalew et al (2012) including the DISC-1 (Disrupted in SZ)gene first identified in a Swedish population of SZ patient Thesegenomic studies were also supported by further proteomic analy-sis that demonstrated differential expression of several proteinsinvolved in neurodevelopment between SZ and normal individ-uals An example of which is the study of Behan et al (2009)of membrane microdomain-associated proteins in dorsolateralprefrontal cortex (DLPFC) The study correlated three proteinswith neurite formation that is a critical step in neurodevel-opment (Freedman 2003) More associations were investigatedat the transcriptomic level as with the study of Hakak et al(2001) using DNA microarray analysis of the gene expression

levels in postmortem DLPFC (Hakak et al 2001) The studydiscovered differential expression of genes implicated in neurode-velopment and synaptic plasticity such as MARCKS GAP-43SCG-10 and neuroserpin (Hakak et al 2001) Moreover theprocess of modeling SZ in terms of different pathways is notcontradictory but rather complimentary Indeed since several ofthese pathways inter-relate and overlap this approach should pro-vide a better insight into the etiology and pathophysiology of thedisease

SYSTEMS BIOLOGY APPLICATION IN BIPOLAR DISORDERAnother application of systems biology in NP disorders involvesthe implication of signal transduction and synaptic plasticity inBPD (Manji and Lenox 2000) Recently a study performed onhuman genetic alterations in SZ and BPD by The SZ Psychi-atric GWAS Consortium showed that the loci of CACNA1CANK3 and ITIH3-4 genes have gained GWS level associationwith BPD (Ripke et al 2011) CACNA1C encodes for the a1Csubunit of the Cav12 voltage-dependent L-type calcium chan-nel (LTCC) which is the major LTCC expressed in mammalianbrain (Bhat et al 2012) It is involved in cell membrane depo-larization and increasing calcium permeability affecting signaltransduction and synaptic plasticity ANK3 encodes a proteinthat is part of the integral membrane proteins associated withaxons ITIH3-4 is involved in extracellular matrix stabilization(Ripke et al 2011) As reviewed by Bhat et al (2012) a single(SNP rs1006737) in CACNA1C gene has been associated withNP diseases in 12 studies associating it with BPD SZ MDD andother psychiatric conditions These genomic results with their rel-evance to the implication of signal transduction and synapticplasticity in BPD are further supported by transcriptomic stud-ies and animal models studied below Moreover CFG approachwas used to analyze candidate genes pathways and mechanismsfor BPD CFG draws upon multiple independent lines of evi-dence for cross-validation of GWAS data to uncover candidategenes and biomarkers involved in disease pathogenesis Theselines of evidence include independent GWAS data animal mod-els and experiments human blood analysis human postmortembrain analysis and human genetic linkage and association studies(Le-Niculescu et al 2009) The study by Ogden et al integratedpharmaco-genomic mouse model that involve treatment of micewith stimulants and mood stabilizers with human data includinglinkage studies and postmortem human brain changes Resultsimplicated several pathways including neurogenesisneurotrophicNT signal transduction circadian synaptic and myelin relatedpathways in the pathogenesis of BPD (Ogden et al 2004) A simi-lar study by Le-Niculescu et al used also CFG to identify candidategenes and mechanisms associated with BPD Thirty-two potentialblood biomarkers were identified The interactions between thesegenes were analyzed independently by Ingenuity Pathway Analy-sis and MetaCore softwares The implicated pathways includedamong others growth factor signaling synaptic signaling celladhesion clock genes and transcription factors (Le-Niculescuet al 2009)

Several transcriptomic studies of gene expression profiles inhumans have confirmed the involvement of glutamate transmis-sion regulation GTPase signaling calciumcalmodulin signaling



and other signal transduction pathways in BPD (Eastwood andHarrison 2000 Nakatani et al 2006 Ginsberg et al 2012) Astudy by Iwamoto et al (2004) carried out a comparison ofgene expression profiles between BPD MDD SZ and controlsubjects in postmortem prefrontal cortices samples using Gene-Spring 50 Results revealed downregulation of genes encodingchannels receptors and transporters Moreoverproteomic studieson postmortem brain tissue of BPD have shown several alter-ations related to synaptic functions these involve reduction ofsynapsin in the hippocampus (Vawter et al 2002) and alteredlevels of synaptosomal associated protein SNAP-25 (Fatemi et al2001) Behan et al (2009) carried out proteomic analysis of mem-brane microdomain-associated proteins in SZ and BPD Resultsdemonstrated several protein alterations in BPD related to synap-tic structure and plasticity (syntaxin-binding protein 1 brainabundant membrane-attached signal protein 1 and others Behanet al 2009) Other proteomic studies implicated similar pro-teins involved in cell signaling and synaptic plasticity such asbrain abundant membrane-attached signal protein 1 prohibitintubulin and others (Beasley et al 2006 Focking et al 2011)

Studies on animal models further support the above-mentioned hypothesis Recently Nanavati et al (2011) using PSDproteome profiles found that mood stabilizers modulate pro-teins related to signaling complexes in rats These proteins includeANK3 that has been implicated in BPD An in vivo study by Duet al has implicated the role of AMPA glutamate receptors inBPD Rats treated with antimanic agents like lithium or valproatereduced hippocampal synaptosomal AMPA receptor subunit glu-tamate receptor 1 (GluR1) levels (Du et al 2004) suggestingthat regulation of glutamatergically mediated synaptic plasticitymay play a role in the treatment of BPD (Du et al 2004) Daoet al has emphasized the role of CACNA1C in BPD Cacna1chaplo-insufficiency was associated with lower exploratory behav-ior decreased response to amphetamine and antidepressant-likebehavior in mice (Dao et al 2010) Taken together these datademonstrate the implication of synaptic plasticity and signaltransduction in the pathophysiology of BPD Further research isrequired in this field as systems biology application to NP diseasesis still in infancy

BIOINFORMATICS IN SYSTEMS BIOLOGYCentral to the practice of systems biology and the understanding ofcomplex biological systems and cellular inter-dependencies is theuse of HT computational techniques which enable investigatorsto figure out the global performance of a system This is a majorchallenge to comprehend inter-dependencies between pathwaysgiven the complexity of biological systems (Mazza et al 2012)The diagram in Figure 2 illustrates the use of computational toolsin systems biology

High-throughput computational techniques take as input alarge set of data points Each data point in Systems Biology researchis a multidimensional vector where m-dimensions describe bio-chemical kinetic and HT-omic features collected from con-ducted experiments and public databases and associated withn-dimensions that describe phenotypic features that act as targetclasses The first target of the computational analysis in systemsbiology is to cluster data points with similar features This is

widely used in systems biology and involves mainly unsuper-vised pattern-recognition or machine-learning algorithms thatcan identify patterns within the data provided by the HT-omictechniques Unsupervised algorithms are pattern discovery algo-rithms that aim to understand the structure of a given data setThe family of unsupervised learning methods includes severaltechniques such as K -means factor analysis principal componentanalysis (PCA) independent component analysis (ICA) hierar-chical and self-organizing maps (SOMs) besides other variantsand extensions (Boutros and Okey 2005) This is in contrast tosupervised learning methods where the training data specifies theclasses to be learned In systems biology such tools will grouptogether genes with similar gene expression patterns metabolitessubject to similar variations or proteins with similar translationpatterns Such pattern discovery is essential to handle the hugedata obtained from DNA microarray and mass spectrometric stud-ies Pattern discovery allows to discover co-expressed genes andco-regulated proteins and also to relate proteomic regulation togenomic regulation Also knowledge based clustering techniquessuch as SOMs model neuronal network to discover data pointswith similar orientations and align them in a connected topologySOMs are artificial neural networks (ANNs) typically trained inan unsupervised fashion to reduce the dimensionality of an inputspace They were instrumental in a study by Noriega (2008) toestablish the absence of central focus in Autism

In the following section we outline the two main steps of the K -means algorithm for a set of observations (x1 x2 xn) whereeach observation is an m-dimensional vector and an initial setK -means (m(0)

1 m(0)2 m(0)

K )Step 1 (Assignment) Each observation is mapped to the cluster

having the closest mean

A(t )i =

xS

∥∥∥xS minusm(t )i

∥∥∥ le ∥∥∥xS minusm(t )j

∥∥∥ forall 1 le j le K

(1)

Step 2 (Update) Set the new means to the centroids of the newformed clusters

m(t+1)i =

1∣∣∣C (t )i

∣∣∣sum

xjisin C (t )i

C (t )i xj (2)

In the above equations t refers to the iteration and i to the clus-ter number Convergence occurs when no change in assignmentsoccurs

More recently ldquosemi-supervisedrdquo clustering algorithms havebeen introduced to systems biology These ontology based clus-tering algorithms can identify gene expression groups in relationto prior knowledge from Gene Ontology (Kang et al 2010) Super-vised clustering techniques are also incorporated to the study ofsystems biology where geneprotein expression data are sortedbased on information from databases involving gene ontologiespathway databases and others into a predictive model (Koteraet al 2012) Supervised clustering algorithms are more involved ingene classification than clustering These algorithms are associatedwith several databases and network clustering software mentionedin what follows

Further involvement of computational methodologies in sys-tems biology includes discovering computational models from



FIGURE 2 | Computational tools in the context of systems biologyProcess of systems biology starting from the HT-discovery tools tillbiomarker discovery and elucidation of biological network results ofHT-discover tools are collected and appropriate validation procedure is runlike western blot Data is then subjected to computational analysispreprocessed to fit into the different algorithms for clustering andclassification These algorithms include supervised semi-supervised andunsupervised statistical and heuristic algorithms along others Knowledgebased and supervisedsemi-supervised algorithms have informationfeeding from public databases that include ontologies signaling andinteraction networks Information from these databases are obtainedthrough appropriate data mining tools and software and fed into thealgorithms These algorithms will give clustered data with an estimatedaccuracy This clustered data is imported to network analysis modelingsimulation and visualization tools These tools get also information from

public databases fed through data mining techniques and also utilize theprevious mentioned algorithms The output of such tools is a model of theimplicated network This model is subject to computational analysis andin silico modeling and simulations for hypotheses screening usingcomputer models These simulations and models allow refining theproposed network and orient the design of appropriate in vitro and in vivoexperiments on a fit animal model or concerned human cells These in vivoand in vitro experiments are run to assess the hypotheses and predictionsof the modeled network and allow for support or refinement of thenetwork After enough evidence is collected through these rounds ofexperimentation and hypotheses testing a final representation of thesystem is devised that could account to a disease pathogenesis or normalphysiology This representation allows for biomarker and new therapeuticdiscovery provides an insight into the pathophysiology or normalphysiology and can help update the online databases

stochastic and probabilistic data points and n-gram analysis withhidden variables using Hidden Markov Models (HMMs) HMMsare stochastic models where the assumed system is a Markovprocess with unobservable states A Markov process can be definedas a stochastic process whose behavior at time t only dependson its behavior at some time t 0 and not times prior to it Thesetechniques subsume almost all other techniques however theyrequire huge computational overhead when analyzing large datasets with dense probabilistic networks Heuristics can be used toreduce the overhead For example the Markov clustering algo-rithm (MCL Bustamam et al 2012) is used in bioinformaticsto cluster proteinndashprotein interaction networks (PPI) and proteinsimilarity networks (Satuluri et al 2010) It is a graph cluster-ing algorithm that relies on probabilistic studies to analyze thenetwork components and the flow within network clusters It

reduces the size of clusters and assigns probabilistic weights forthe components The MLC algorithm does not perform well withlarge data sets and forms a large number of imbalanced small clus-ters Therefore Satuluri et al (2010) used well known facts aboutthe structure of PPI networks to regulate MCL and introduced theMulti-Level-Regularized-MCL (MLR-MCL)

In addition to clustering methods computational toolsinvolved in systems biology include the use of distance metricsthat can vary from Euclidean to information based distance met-rics Distance metrics can help detect similarities of genes or otherobtained samples These metrics are among the primary strategiesused in microarray analysis but have certain drawbacks For exam-ple the main limitation of Euclidean distance (also referred toas the l2 norm) resides in its sensitivity to any outliers in thedata Alternatives include using measures of statistical dependence



The simplest measure of dependence is Pearsonrsquos correlation It iswidely used and is able to capture linear dependence betweentwo variables Other measures of correlation like Spearmanrsquos rhoand Kendallrsquos tau are also popular but can only capture monot-one dependence Non-linear dependence can be measured usingmutual information or other measures of association (Reacutenyi1959) In general the choice of a measure of dependence is gov-erned by the availability of a solid estimator and a relatively lowcomputational burden A popular rank-based measure of concor-dance between variables Kendallrsquos tau has been used in severalsystems biology contexts (Bolboaca and Jantschi 2006 Sen 2008)and can be defined as follows

τ = 2Ncp minus Nncp

n (n minus 1)(3)

where in Eq 3 N cp and N ncp respectively refer to the number ofconcordant or discordant pairs where for a set of observations(x1 y1) (xn yn)

A concordant pair is a pair where xi lt xj and yi lt yj or xi gt xj andyi gt yj

A discordant pair is a pair where xi lt xj and yi gt yj or xi gt xj andyi lt yj

Other computational methodologies have been also used in relat-ing data points to existing public databases using data miningtechniques such as association rules decision trees and patternbased searches This can result in updating the public databases orin refining the experiments

The above-mentioned algorithms and strategies have been usedby several software libraries and programs available for use in thedatabase data mining and network clustering modeling and sim-ulation areas (Table 1) Examples of these resources in the databaseand data mining areas are Genomics (KEGG Kanehisa et al2012 Human Gene Expression Index Haverty et al 2002 andTRED Jiang et al 2007) proteomics (GELBANK Babnigg andGiometti 2004 X Tandem Craig et al 2004 and PRIDE Jonesand Cote 2008) signaling pathways (PANTHER Mi and Thomas

Table 1 | Examples of bioinformatic resources and computational tools

Database URL Reference

DATA RESOURCES DATABASES FOR ONTOLOGIES GENOMICS PROTEOMICS SIGNALING PATHWAYS AND INTERACTION NETWORKS

KEGG httpwwwgenomejpkegg Kanehisa et al (2012)

Human Gene Expression Index httpwwwbiotechnologycenterorahio Haverty et al (2002)

TRED httprulaicshleducgi-binTREDtredcgiprocess=home Jiang et al (2007)

GELBANK httpgelbankanlgov Babnigg and Giometti (2004)

X Tandem httpwwwtheapmorqTANDEM Craig et al (2004)

PRIDE ftpftpebiacukpubdatabasespride Jones and Cote (2008)

PANTHER httpwwwpantherdbora Mi and Thomas (2009)

GenMAPP2 httpwwwaenmappora Salomonis et al (2007)

Reactome httpwwwreactomeorg Croft et al (2011)

TRANSPATH httpwwwaenexplaincomtranspath Krull et al (2006)

IntAct httpwwwebiacukintactindexisp Kerrien et al (2012)

MIPSMPPI httpmipsgsfdeprojppi Pagel et al (2005)

aMAZE httpwwwamazeulbacbe Lemer et al (2004)

GeneNet httpwwwmgsbionetnscrumgsgnwqenenet Ananko et al (2002)

GO ndash Gene Ontology httpwwwgeneontoloavoraGOdownloadsshtml Harris et al (2004)

SO ndash Sequence Ontology httpobosourceforgenetcgi-bindetailcgiseauence

PhenomicDB httpwwwphenomicdbde Kahraman et al (2005)

DATA ANALYSISTOOLS NETWORK ANALYSIS SIMULATION ANDOR MODEL ASSESSMENT

MATLAB Simulink toolbox httpwwwmathworkscomproductssimulink Ullah et al (2006)

Virtual Cell httpwwwnrcamuchcedu Moraru et al (2002)

JWS online httpiiibiochemsunaczaindexhtml Olivier and Snoep (2004)

Ingenuity Pathway Analysis httpwwwingenuitycomproductspathways_analysishtml Jimenez-Marin et al (2009)

NetBuilder httphomepaqesstcahertsacuk~erdqmjsNetBuilder20homeNetBuilder

Copasi httpwwwcopasiora Mendes et al (2009)

E-cell httpwwwe-cellorg Takahashi et al (2003)

Cell Designer httpwwwcelldesianerora Van Hemert and Dickerson (2010)

Cellware httpwwwbiia-staredusqresearchsbqcellwareindexasp Dhar et al (2004)

SimCell httpwishartbiologyualbertacaSimCell Tretter and Gebicke-Haerter (2012)

NETWORK VISUALIZATIONTOOLS

Cytoscape httpwwwcytoscapeorg Shannon et al (2003)

BioLayout httpwwwbiolayoutorg Theocharidis et al (2009)

Cobweb httpbioinformaticscharitedecobweb cobweb von Eichborn et al (2011)



2009 GenMAPP2 Salomonis et al 2007 Reactome Croft et al2011 and TRANSPATH Krull et al 2006) and interaction net-works (IntAct Kerrien et al 2012 MIPSMPPI Pagel et al 2005and aMAZE Lemer et al 2004)

Other computational tools are involved in network visual-ization analysis simulation andor model assessment Theseinclude MATLABSimulink toolbox (Ullah et al 2006) V-Cell(Moraru et al 2002) JWS online (Olivier and Snoep 2004)Copasi (Mendes et al 2009) E-cell (Takahashi et al 2003)Ingenuity Pathway Analysis (Jimenez-Marin et al 2009) CellDesigner (Van Hemert and Dickerson 2010) Cellware (Dharet al 2004) Cytoscape with the Network Analyzer Plugin (Shan-non et al 2003) and other tools like Simcell for more com-plex automated simulations (Tretter and Gebicke-Haerter 2012)Some tools emphasize only network visualization like BioLay-out (Theocharidis et al 2009) and Cobweb (von Eichborn et al2011) An important tool used in analysis and visualization ofHT-omics data is Bioconductor (Gentleman et al 2004) Biocon-ductor is a tool based on ldquoRrdquo ldquoRrdquo is statistical package used inseveral tools and softwares of systems biology It is an environ-ment dedicated for statistical computing and graphics involvinglinear and non-linear modeling classification and clustering ofthe obtained data Its application through Bioconductor involvesseveral packages like HTqPCR for HT quantitative real-time PCRassays MassSpecWavelet and PROcess for mass spectrometricdata processing VegaMC for comparative genome hybridizationdatasets (Morganella and Ceccarelli 2012) easyRNAseq for RNAsequence datasets (Delhomme et al 2012) RedeR (Castro et al2012) and graphite (Sales et al 2012) for biological networksamongst others

These environments algorithms software and databases con-stitute the core of action of systems biology in its attempt todecipher the complexity of biological systems Furthermore theemergence of commodity multi-core and concurrent processingtechnology with Graphical Processing Units (GPUs) and the algo-rithmic advances and discovery of novel more efficient algorithmsenable novel and practical applications of systems biology

APPLICATION OF BIOINFORMATICS AND IN SILICOMODELING IN NEUROSCIENCE AND NEUROPSYCHIATRYAs NP disorders are categorized among the highest complex dis-eases they have been subject for advanced computational tools formodeling simulation and network analysis Predictive systemsbiology has been applied to the analysis of SZ where several in sil-ico models and several simulations have been proposed to drawinferences on the relation between biophysical alteration and cor-tical functions as it will be illustrated in this section (Han et al2003 Qi et al 2008 2010 Vierling-Claassen et al 2008 Spencer2009 Hoffman et al 2011 Morris et al 2011 Rotaru et al 2011Volman et al 2011 Waltz et al 2011 Cano-Colino and Compte2012 Komek et al 2012) Computational modeling may help tobridge the gaps between postmortem studies animal models andexperimental data in humans (Spencer 2009) One major appli-cation of modeling is related to gamma band oscillations that arefound to be altered in case of SZ and are responsible for sev-eral alterations and symptoms Gamma activity in people with SZappears to have less amplitude and less synchronization

One study examining the relationship between schizophrenicsymptom profile and oscillatory activity in the gamma range haveidentified that positive symptoms such as reality distortion orhallucinatory activity associated with increased gamma activitywhile negative symptoms such as psychomotor poverty are asso-ciated with decreased gamma oscillations in human brain circuitry(Vierling-Claassen et al 2008) In this study two computationalmodels illustrating gamma band changes in SZ patients were sug-gested based on previous studies demonstrating that SZ expresseddecreased GAT-1 (GABA transporter) and GAD67 (responsible forGABA synthesis) Based on experimental evidence the blockadeof GAT-1 has been modeled as extended inhibitory pos-tsynapticcurrent (IPSC) whereas GAD67 decrease can be modeled as adecrease in strength of inhibition (Vierling-Claassen et al 2008)

A recent study done by Komek et al (2012) modeled theeffect of dopamine on GABA-ergic transmission and Schizo-phrenic phenotype Through its action on GABA-ergic neuronsdopamine increases the excitability of fast-spiking interneuronsThe effect of dopamine was demonstrated through varying leakK+ conductance of the fast-spiking interneurons and gamma bandoscillations Results of the simulations indicate that dopaminecan modulate cortical gamma band synchrony in an inverted-Ufashion and that the physiologic effects of dopamine on single fast-spiking interneurons can give rise to such non-monotonic effectsat the network level as illustrated in Figure 3 Figure 3 illustrateshow amphetamine administration increases gamma synchrony inSZ patients but decreases it in controls Patients lie on the leftside of the curve and controls at the optimal point (Komek et al2012) Moreover studying several impairments associated withSZ such as working memory shows that these impairments followthe same curve (Komek et al 2012 Tretter and Gebicke-Haerter2012) This model allows for predicting-based on the initial posi-tion on the curve-the effect that dopaminergic drugs can have oncognitive function (Cools and DrsquoEsposito 2011) This inverted-Uassociation between dopaminergic stimulation and prefrontal cor-tex activity has been previously reported in the literature by severalstudies including those conducted by Goldman-Rakic et al (2000)and Seamans and Yang (2004)

Furthermore Spencer simulated a model of human corticalcircuitry using 1000 leaky integrate-and-fire neurons that can sim-ulate neuronal synaptic gamma frequency interactions betweenpyramidal neurons and interneurons (for a review about the useof integrate-and-fire neurons in brain simulation check (Burkitt2006) Spencer (2009) studied the effect of varying several synap-tic circuitry components on the gamma band oscillation andnetwork response and proposed that a multimodal approachcombining non-invasive neurophysiological and structural mea-sures might be able to distinguish between different neural circuitabnormalities in SZ patients Volman and colleagues modeled Par-valbumin (PV)-expressing fast-spiking interneurons that interactwith pyramidal cells (PCs) resulting in gamma band oscillationsto study the effect of PV and GAD67 on neuronal activity Thismodel suggested a mechanism by which reduced GAD67 and PVin fast-spiking interneurons may contribute to cortical dysfunc-tion in SZ (Volman et al 2011) The pathophysiology behindNMDA hypofunction in SZ has been suggested by a model pro-posed by Rotaru et al (2011) Modeling a network of fast-spike



FIGURE 3 | Predictive model of dopamine stimulation Inverted-U graphof prefrontal cortex function depending on the level of dopaminestimulation Dopamine agonists have a therapeutic effect on Schizophreniathat is a state of reduced dopamine stimulation while these agonists causeadverse effects to controls who have optimal dopamine stimulation(Adapted with modifications from Komek et al 2012)

(FS) neurons and PCs they found that brief α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR)-mediatedFS neuron activation is crucial to synchronize PCs in the gammafrequency band

Neuro-computational techniques have been also employedto detect possible mechanisms for several endophenotypes andimpairments in SZ For example the model by Waltz et al (2011)correlates deficit in procedural ldquoGordquo learning (choosing the beststimulus at a test) with dopamine alterations Another model byMorris et al (2011) suggests an association between weakenedrepresentation of response values and the failure to associate stim-uli with appropriate response alternatives by the basal gangliaOther models associate abnormal connectivity or NMDA recep-tor dysfunction with SZ features including deficits of simple spatialworking memory (Cano-Colino and Compte 2012) associativememory recall (Han et al 2003) and narrative language dis-ruption (Hoffman et al 2011) Therefore these computationalmodels have also showed how abnormal cortical connectivity andsynaptic abnormalities concerning NMDA receptors and gluta-minergic transmission tend to explain some of the endopheno-types of SZ These findings fortify the association between thesephysiological changes and overt phenotype of SZ and provide acommon way to study the etiological basis of several NP disordersendophenotypes

Models of dopamine homeostasis have been demonstratedgain insight into the actual mechanisms underlying the dopaminetransmission its role in SZ and other NP disorders and the impli-cations on therapy and disease management Qi et al (2008)reviewed the hypotheses related to dopamine homeostasis andrevised them in the context of SZ using a proposed neuro-computational model Another mathematical model proposed bythe same author included a biochemical system theory that mod-els a system of relevant metabolites enzymes transporters and

regulators involved in the control of the biochemical environmentwithin the dopamine neuron to assess several components andfactors that have been implicated in SZ Such a model is proposedto act as a screening tool for the effect of dopamine therapiesameliorating the symptoms of SZ (Qi et al 2008)

Taken together these models (dopamine homeostasis andgamma band synchrony along with other proposed models) pro-vide some aspect of the predictive part of systems biology Theyprovide an insight on how mechanisms could predict the roles ofproteomic findings in the determination of the overt phenotypeBy this they can complement hypotheses and results presentedby the integrative branch mentioned before For example neuro-computational modeling of cortical networks by Volman et al(2011) has provided one mechanism by which GAD67 reductionhelps in understanding the pathophysiology of SZ and demon-strates clearly how systems biology ndash through both the integra-tive and predictive faces ndash has the unique potential to dig intothe underlying mechanisms involved in the perturbed biologicalsystem under certain disease condition

Another similar application of NP computational modelinginvolves autism Vattikuti and Chow (2010) analyzing the mech-anisms linking synaptic perturbations to cognitive changes inautism showed that hypometria and dysmetria are associated withan increase in synaptic excitation over synaptic inhibition in cor-tical neurons This can be due to either reduced inhibition orincreased excitation at the level of the different synapses (Vattikutiand Chow 2010) This study draws inferences about the possi-ble link between perturbation of cerebral cortical function andautistic phenotype and demonstrates that different varied phar-macological approaches should be used in treating the autismsymptoms Other computational models used SOMs algorithmsto study sensory abnormalities in autism (Noriega 2008) A studyby Noriega supported the notion that weak central coherence isresponsible for sensory abnormalities in autism These findingssuggest that failure in controlling natural variations in sensitivi-ties to sensory inputs could be referred back to faulty or absentfeedback mechanism (Noriega 2008)

A previous study by Noriega (2007) uses the same SOM-algorithm to study abnormalities in neural growth in the brainsof autistic children and sensory abnormalities associated with theautism in these children Neural growth abnormalities featureswere compared to the effects of manipulating physical structureand size of these SOM Results did not show an impact of theabnormalities on stimuli coverage but a negative effect on mapunfolding Sensory abnormalities were studied through atten-tion functions that can model hypersensitivity and hyposensitivity(Noriega 2007) The model successfully reproduced the potentialof autistic patients to focus on details rather than the whole andproved no connection between noisy neuronal communication inthese individuals and the autistic phenotype Moreover hypoth-esizing that the key to understanding mechanisms of autism dis-orders relies in elucidating the perturbations affecting synaptictransmission a computational model of synaptic transmissionhas been investigated In a study by Gowthaman et al (2006)a computational model of the snapin protein and the SNAREcomplex interaction which is involved in NT release have beenpresented This model allows for a better understanding of the



role of snapin in SNARE regulation thus in NT release Thesemodels may have further implications on therapeutic discoverySeveral other computational models have been devised to ana-lyze various symptoms and endophenotypes of autism (Neumannet al 2006 Triesch et al 2006) Two independent studies byTriesch et al (2006) and Neumann et al (2006) proposed com-putational models of the emergence of gaze in children that canpredict possible processes underlying the gaze abnormality asso-ciated with behavioral impairment in autism Moreover modelsof cortical connectivity by Just et al (2012) and Kana et al (2011)relate cortical under-connectivity between frontal and posteriorcortex to the impaired ability of autistic patients to accomplishcomplex cognitive and social tasks thus suggesting an expla-nation for cognitive and behavioral impairments in autism Inconclusion these models provide a sort of computational assayof symptoms of autism and support the association of thesesymptoms with the underlying cortical connectivity and synapticalterations

BIOMARKERS AND THEIR USE IN PSYCHIATRYBiomarkers in psychiatry and application of HT technology arestill in infancy with very limited clinically useful markers (Les-cuyer et al 2007) A spike in biomarker research has occurredduring the last 10 years due to different applications and advan-tages which have been applied to the field of neuroscience (Woodset al 2012) As one molecular biomarker alone may not have astrong statistical power to predict outcomes especially with com-plex diseases the current trend is using HT and systems biologytechniques to identify a set of biomarkers or surrogate markersthat can be used as a panel to characterize a certain disorder ordisease This has been a continual major challenge in biomarkerdiscovery (Biomarkers Definitions Working Group 2001 Woodset al 2012)

Biomarkers can be used as surrogate endpoints Surrogateendpoints are biomarkers that intended to substitute a clinicalendpoint (Biomarkers Definitions Working Group 2001 Filiouand Turck 2012) Such biomarkers have been widely sought inNP diseases especially in the body fluid A perfect marker shouldbe easily accessible in a non-invasive manner Therefore bloodhas been considered ideal as a source for the quest for biomarkers(Lescuyer et al 2007 Dao et al 2010) In addition it reflects theentire homeostasis of the body and contacts every organ

However several difficulties are associated with the identifica-tion process of clinically useful serum biomarkers including thelarge amount of proteins present with their high dynamic rangeof abundance that can span 10ndash12 orders of magnitude Such alevel cannot be reached even by the advanced HT techniques (Daoet al 2010 Korolainen et al 2010 Juhasz et al 2011) Therebythis causes high abundance proteins to mask lower abundant ones(Juhasz et al 2011) especially considering that only 22 plasma pro-teins constitute 99 of the total plasma proteins (Lescuyer et al2007) Other limitations include the presence of diverse analytesincluding proteins small lipids electrolytes in the serum (Zhanget al 2005 Dao et al 2010) Another candidate biofluid forbiomarker investigation is the cerebrospinal fluid (CSF) BecauseCSF is in communication with cerebral extracellular fluid and isless hampered by confounding factors therefore more accurately

reflect cerebral pathological changes CSF seems to be the mostpromising source for both ND and NP diseases (Zhang et al2005) Further advantages of CSF investigation include higher con-centrations especially that several molecules cannot cross into theplasma due to the blood brain barrier In addition CSF reflects themetabolic processes occurring within the brain (Jiang et al 2003Korolainen et al 2010 Juhasz et al 2011) However problemscould be associated with CSF sampling as many still consider lum-bar puncture as an invasive procedure (Jiang et al 2003 Zhanget al 2005 Juhasz et al 2011)

Biomarkers in NP can aid in staging and classification of theextent of a disease (Biomarkers Definitions Working Group 2001Domenici et al 2010) allowing for disease stratification which isthe basis of personalized medicine (Zhang et al 2010) Further-more biomarker discovery in the field of NP diseases allows forthe discovery of new pathways relevant to the pathophysiology ofthat particular disorder For example miRNA 173 and its targetsTCF4 and CACNA1C were recognized as susceptibility genes inSZ through GWAS studies previously mentioned This discoveryallowed for the implication of abnormal neuronal development(neurogenesis) into the etiology of SZ since miRNA 173 is a knownregulator of human neurogenesis (Ripke et al 2011)

Diagnosis and treatment of patients with SZ is another exam-ple of clinical challenge in psychiatry Currently there is notenough understanding about pathophysiology of SZ and pharma-cotherapy This usually leads to multiple trials evaluating differentantipsychotics until desirable clinical effects are achieved (Schwarzet al 2012) In an attempt to gain insights into pathophysiologyof SZ Cai et al used a mass spectrometry-based metabolomicanalysis to evaluate altered metabolites in SZ patients In theirwork they compared metabolic monoamine and amino acid NTmetabolites in plasma and urine simultaneously between first-episode neuroleptic-naive SZ patients and healthy controls beforeand after a 6-weeks risperidone monotherapy Their findingsshow that antipsychotic treatment (risperidone) can approachNT profile of patients with SZ to normal levels suggesting thatrestoration of the NT may be parallel with improvement in psy-chotic symptoms They also used LC-MS and H nuclear magneticresonance-based metabolic profiling to detect potential markersof SZ They identified 32 candidates including pregnanediol cit-rate and α-ketoglutarate (Cai et al 2012) Further studies withlarger numbers of patients will be required to validate these find-ings and to determine whether these markers can be translatedinto clinically useful tests So far no biomarker has been approvedfor clinical use in SZ (Macaluso and Preskorn 2012)

As for drug discovery application biomarkers identification canhave a direct application helping to identify new treatment targetsFor example the recognition of DNA methyltransferase (DNMT)inhibitors has been shown as a potential candidate for the treat-ment of SZ studied on mouse model (Satta et al 2008) Recentstudies have demonstrated the effects of DNMTs on hypermethy-lation and downregulation of GAD67 a novel potential biomarkerfor SZ (Ptak and Petronis 2010)

Finally clinically validated biomarkers would aid physicians inearly diagnosis and would allow a better and more rigid discrim-ination between different NP diseases (Biomarkers DefinitionsWorking Group 2001) Several studies have been established to



discover biomarkers that can draw the line between different NPdiseases like MDD and BPD (Ginsberg et al 2012) SZ and BPD(Thomas et al 2003 Harris et al 2007 Glatt et al 2009) SZ anddepression (Domenici et al 2010) Biomarker discovery can alsoassist in determining the course of the disorder and when and howto treat (Woods et al 2012)

CONCLUSIONNeuropsychiatric disorders including SZ major depression BPDsare difficult to tackle and track down since their occurrenceinvolves several genes several epigenetic mechanisms and envi-ronmental effects Therefore the emergence of systems biol-ogy as a discipline provides a mean to investigate the patho-physiology of these NP disorders using a global and system-atic approach Systems biology allows a thorough investigationof the system components its dynamics and responses to anykind of perturbations at the base line level and the experimen-tal level It integrates mathematical models with experimentalmolecular information from in silico in vivo and in vitro stud-ies In SZ the application of systems biology underlined thedeleterious effects of GABA that disrupt both the glutaminergicand dopaminergic transmission When applied in BPD systemsbiology revealed the implication of a number of genes includ-ing interactions of CACNA1C ANK3 and ITIH3-4 genes viagenetic association analysis In autism systems biology high-lighted a link between synaptic perturbations and cognitiveimpairments

The systems biology discipline makes use of available or novelbioinformatics resources and computational tools to model bio-logical systems in question It enables a better comprehension ofthe inter-dependencies between pathways and the complexity ofbiological systems It provides simulation and prediction abilitiesto investigate how systems react upon interferences

Along the same lines systems biology techniques requiresignificant data preprocessing to overcome a number of keychallenges First there are challenges pertaining to the bias instatistical analysis and the difficulties in performing clinical val-idation (Frank and Fossella 2011 Linden 2012 Villoslada andBaranzini 2012) Second many studies are confined to a smallsample size that hinders the ability to infer conclusions of high con-fidence (Bhat et al 2012) Third theldquoomicsrdquodiscovery approachessuffer from the heterogeneity of the sampled populations whichinvolves interspecies and genetic variations (Cowan et al 2002OrsquoTuathaigh et al 2012) Such heterogeneity could account inpart for the conflicting results associated with the limited experi-mental reproducibility (Fathi et al 2009 Martins-de-Souza et al2012) Furthermore as most studies rely on postmortem humantissue challenges encompass the inability to find enough sam-ples (Korolainen et al 2010 Sequeira et al 2012) along with theinherent limitations of neuropathological analysis to discriminatebetween changes caused by the NP disorders themselves and thosederived from postmortem artifacts or other confounding factors(Huang et al 2006 Harris et al 2007) Furthermore most ofthese studies are related to late-onset stages of the disease andonly in few cases researchers are able to reflect early changes thatcharacterize these NP disorders (Juhasz et al 2011)

Systems biology in the areas of psychiatry and neurosciencehas provided limited insights into the pathophysiology and themolecular pathogenesis of NP diseases For all of these studiescomputational techniques used in systems biology are consid-ered necessary tools to determine functional associations of keyfactors and pathways involved in NP disorders Finally the pro-jected utility of systems biology will be valuable for discoveringdisease-specific NP biomarkers Such markers are key tools fordisease assessment and for the discovery of novel treatments inneuropsychiatry

REFERENCESAbdolmaleky H M Cheng K H

Faraone S V Wilcox M Glatt SJ Gao F et al (2006) Hypomethy-lation of MB-COMT promoter isa major risk factor for schizophre-nia and bipolar disorder Hum MolGenet 15 3132ndash3145

Abraham J E Maranian M J SpiteriI Russell R Ingle S Luccarini Cet al (2012) Saliva samples are aviable alternative to blood samples asa source of DNA for high throughputgenotyping BMC Med Genomics519 doi1011861755-8794-5-19

Allen N C Bagade S McqueenM B Ioannidis J P KavvouraF K Khoury M J et al(2008) Systematic meta-analysesand field synopsis of genetic asso-ciation studies in schizophrenia theSzGene database Nat Genet 40827ndash834

Altamura A C Pozzoli S FiorentiniA and DellrsquoOsso B (2012) Neu-rodevelopment and inflammatorypatterns in schizophrenia in

relation to pathophysiologyProg Neuropsychopharmacol BiolPsychiatry doi 101016jpnpbp201208015 [Epub ahead of print]

Ananko E A Podkolodny N L Stepa-nenko I L Ignatieva E V Pod-kolodnaya O A and Kolchanov NA (2002) GeneNet a database onstructure and functional organisa-tion of gene networks Nucleic AcidsRes 30 398ndash401

Archer T (2010) Neurodegenerationin schizophrenia Expert Rev Neu-rother 10 1131ndash1141

Arnold S E Talbot K and HahnC G (2005) Neurodevelopmentneuroplasticity and new genes forschizophrenia Prog Brain Res 147319ndash345

Ayalew M Le-Niculescu H LeveyD F Jain N Changala BPatel S D et al (2012) Con-vergent functional genomics ofschizophrenia from comprehen-sive understanding to genetic riskprediction Mol Psychiatry 17887ndash905

Babnigg G and Giometti C S (2004)GELBANK a database of annotatedtwo-dimensional gel electrophoresispatterns of biological systems withcompleted genomes Nucleic AcidsRes 32 D582ndashD585

Baumeister A A and Francis J L(2002) Historical development ofthe dopamine hypothesis of schiz-ophrenia J Hist Neurosci 11265ndash277

Beasley C L Pennington K BehanA Wait R Dunn M J and Cot-ter D (2006) Proteomic analy-sis of the anterior cingulate cor-tex in the major psychiatric disor-ders evidence for disease-associatedchanges Proteomics 6 3414ndash3425

Behan A T Byrne C Dunn M JCagney G and Cotter D R (2009)Proteomic analysis of membranemicrodomain-associated proteins inthe dorsolateral prefrontal cortex inschizophrenia and bipolar disorderreveals alterations in LAMP STXBP1and BASP1 protein expression MolPsychiatry 14 601ndash613

Bhat S Dao D T Terrillion C EArad M Smith R J SoldatovN M et al (2012) CACNA1C(Ca(v)12) in the pathophysiology ofpsychiatric disease Prog Neurobiol99 1ndash14

Biomarkers Definitions WorkingGroup (2001) Biomarkers andsurrogate endpoints preferreddefinitions and conceptual frame-work Clin Pharmacol Ther 6989ndash95

Bolboaca S D and Jantschi L (2006)Pearson versus Spearman Kendallrsquostau correlation analysis on structure-activity relationships of biologicactive compounds Leonardo J Sci5 179ndash200

Boutros P C and Okey A B (2005)Unsupervised pattern recognitionan introduction to the whys andwherefores of clustering microarraydata Brief Bioinform 6 331ndash343

Burkitt A N (2006) A review of theintegrate-and-fire neuron model IHomogeneous synaptic input BiolCybern 95 1ndash19



Bustamam A Burrage K and Hamil-ton N A (2012) Fast parallelMarkov clustering in bioinformaticsusing massively parallel computingon GPU with CUDA and ELLPACK-R sparse format IEEEACM TransComput Biol Bioinform 9 679ndash692

Cacabelos R Hashimoto R andTakeda M (2011) Pharmacoge-nomics of antipsychotics efficacy forschizophrenia Psychiatry Clin Neu-rosci 65 3ndash19

Cai H L Li H D Yan X ZSun B Zhang Q Yan M et al(2012) Metabolomic analysis of bio-chemical changes in the plasma andurine of first-episode neuroleptic-naive schizophrenia patients aftertreatment with risperidone J Pro-teome Res 11 4338ndash4350

Cano-Colino M and Compte A(2012) A computational model forspatial working memory deficits inschizophrenia Pharmacopsychiatry45(Suppl 1) S49ndashS56

Castro M A Wang X Fletcher MN Meyer K B and MarkowetzF (2012) RedeR RBioconductorpackage for representing modu-lar structures nested networks andmultiple levels of hierarchical asso-ciations Genome Biol 13 R29

Clark D Dedova I Cordwell S andMatsumoto I (2006) A proteomeanalysis of the anterior cingulate cor-tex gray matter in schizophreniaMol Psychiatry 11 459ndash470 423

Cools R and DrsquoEsposito M (2011)Inverted-U-shaped dopamineactions on human working mem-ory and cognitive control BiolPsychiatry 69 e113ndashe125

Cowan W M Kopnisky K L andHyman S E (2002) The humangenome project and its impact onpsychiatry Annu Rev Neurosci 251ndash50

Coyle J T (2006) Glutamate and schiz-ophrenia beyond the dopaminehypothesis Cell Mol Neurobiol 26365ndash384

Craig R Cortens J P and Beavis R C(2004) Open source system for ana-lyzing validating and storing pro-tein identification data J ProteomeRes 3 1234ndash1242

Crespi B Stead P and Elliot M(2010) Evolution in health andmedicine Sackler colloquium com-parative genomics of autism andschizophrenia Proc Natl Acad SciUSA 107(Suppl 1) 1736ndash1741

Croft D OrsquoKelly G Wu G HawR Gillespie M Matthews L etal (2011) Reactome a databaseof reactions pathways and biologi-cal processes Nucleic Acids Res 39D691ndashD697

da Silva Alves F Figee M vanAvamelsvoort T Veltman D andDe Haan L (2008) The reviseddopamine hypothesis of schizophre-nia evidence from pharmacologicalMRI studies with atypical antipsy-chotic medication Psychopharma-col Bull 41 121ndash132

Dao D T Mahon P B Cai X Kovac-sics C E Blackwell R A AradM et al (2010) Mood disorder sus-ceptibility gene CACNA1C modifiesmood-related behaviors in mice andinteracts with sex to influence behav-ior in mice and diagnosis in humansBiol Psychiatry 68 801ndash810

Delhomme N Padioleau I FurlongE E and Steinmetz L M (2012)easyRNASeq a bioconductor pack-age for processing RNA-Seq dataBioinformatics 28 2532ndash2533

Dhar P Meng T C Somani SYe L Sairam A Chitre M etal (2004) Cellware ndash a multi-algorithmic software for computa-tional systems biology Bioinformat-ics 20 1319ndash1321

Domenici E Wille D R Tozzi FProkopenko I Miller S Mck-eown A et al (2010) Plasmaprotein biomarkers for depres-sion and schizophrenia by multianalyte profiling of case-controlcollections PLoS ONE 5e9166doi101371journalpone0009166

Du J Quiroz J Yuan P Zarate Cand Manji H K (2004) Bipolar dis-order involvement of signaling cas-cades and AMPA receptor traffick-ing at synapses Neuron Glia Biol 1231ndash243

Eastwood S L and Harrison PJ (2000) Hippocampal synapticpathology in schizophrenia bipo-lar disorder and major depressiona study of complexin mRNAs MolPsychiatry 5 425ndash432

Egerton A and Stone J M (2012) Theglutamate hypothesis of schizophre-nia neuroimaging and drug devel-opment Curr Pharm Biotechnol 131500ndash1512

Enomoto T Noda Y and NabeshimaT (2007) Phencyclidine and geneticanimal models of schizophreniadeveloped in relation to the gluta-mate hypothesis Methods Find ExpClin Pharmacol 29 291ndash301

Fang F C and Casadevall A (2011)Reductionistic and holistic scienceInfect Immun 79 1401ndash1404

Fatemi S H Earle J A Stary J M LeeS and Sedgewick J (2001) Alteredlevels of the synaptosomal associ-ated protein SNAP-25 in hippocam-pus of subjects with mood disordersand schizophrenia Neuroreport 123257ndash3262

Fathi A Pakzad M Taei A Brink TC Pirhaji L Ruiz G et al (2009)Comparative proteome and tran-scriptome analyses of embryonicstem cells during embryoid body-based differentiation Proteomics 94859ndash4870

Filiou M D and Turck C W(2012) Psychiatric disorder bio-marker discovery using quantitativeproteomics Methods Mol Biol 829531ndash539

Focking M Dicker P English J ASchubert K O Dunn M J andCotter D R (2011) Common pro-teomic changes in the hippocam-pus in schizophrenia and bipolardisorder and particular evidencefor involvement of cornu ammonisregions 2 and 3 Arch Gen Psychiatry68 477ndash488

Food and Drug Administration HH S (2011) International Confer-ence on Harmonisation Guidanceon E16 biomarkers related to drugor biotechnology product develop-ment context structure and for-mat of qualification submissionsavailability Notice Fed Regist 7649773ndash49774

Frank M J and Fossella J A (2011)Neurogenetics and pharmacologyof learning motivation and cogni-tion Neuropsychopharmacology 36133ndash152

Freedman R (2003) Schizophrenia NEngl J Med 349 1738ndash1749

Gentleman R C Carey V J Bates DM Bolstad B Dettling M DudoitS et al (2004) Bioconductor opensoftware development for computa-tional biology and bioinformaticsGenome Biol 5 R80

Ginsberg S D Hemby S E andSmiley J F (2012) Expressionprofiling in neuropsychiatric disor-ders emphasis on glutamate recep-tors in bipolar disorder PharmacolBiochem Behav 100 705ndash711

Glatt S J Chandler S D Bousman CA Chana G Lucero G R TatroE et al (2009) Alternatively splicedgenes as biomarkers for schizophre-nia bipolar disorder and psychosisa blood-based spliceome-profilingexploratory study Curr Pharma-cogenomics Person Med 7 164ndash188

Goldman-Rakic P S Muly E C IIIand Williams G V (2000) D(1)receptors in prefrontal cells and cir-cuits Brain Res Brain Res Rev 31295ndash301

Gormanns P Mueller N S Ditzen CWolf S Holsboer F and Turck CW (2011) Phenome-transcriptomecorrelation unravels anxiety anddepression related pathways J Psy-chiatr Res 45 973ndash979

Gowthaman R Silvester A J SaranyaK Kanya K S and Archana NR (2006) Modeling of the poten-tial coiled-coil structure of snapinprotein and its interaction withSNARE complex Bioinformation 1269ndash275

Greenwood T A Lazzeroni L C Mur-ray S S Cadenhead K S CalkinsM E Dobie D J et al (2011)Analysis of 94 candidate genes and12 endophenotypes for schizophre-nia from the Consortium on theGenetics of Schizophrenia Am JPsychiatry 168 930ndash946

Gustavsson A Svensson M Jacobi FAllgulander C Alonso J Beghi Eet al (2011) Cost of disorders of thebrain in Europe 2010 Eur Neuropsy-chopharmacol 21 718ndash779

Hakak Y Walker J R Li C WongW H Davis K L Buxbaum J Det al (2001) Genome-wide expres-sion analysis reveals dysregulation ofmyelination-related genes in chronicschizophrenia Proc Natl Acad SciUSA 98 4746ndash4751

Han S D Nestor P G Shenton M ENiznikiewicz M Hannah G andMccarley R W (2003) Associativememory in chronic schizophreniaa computational model SchizophrRes 61 255ndash263

Harris M A Clark J Ireland ALomax J Ashburner M FoulgerR et al (2004) The Gene Ontol-ogy (GO) database and informat-ics resource Nucleic Acids Res 32D258ndashD261

Harris L J W Swatton J E Wengen-roth MWayland M Lockstone HE Holland A et al (2007) Differ-ences in protein profiles in schizo-phrenia prefrontal cortex comparedto other major brain disorders ClinSchizophr Relat Psychoses 1 73ndash91

Haverty P M Weng Z Best N LAuerbach K R Hsiao L L JensenR V et al (2002) HugeIndex adatabase with visualization tools forhigh-density oligonucleotide arraydata from normal human tissuesNucleic Acids Res 30 214ndash217

Hayashi-Takagi A and Sawa A(2010) Disturbed synaptic connec-tivity in schizophrenia convergenceof genetic risk factors during neu-rodevelopment Brain Res Bull 83140ndash146

Hoffman R E Grasemann U Gue-orguieva R Quinlan D Lane Dand Miikkulainen R (2011) Usingcomputational patients to evaluateillness mechanisms in schizophre-nia Biol Psychiatry 69 997ndash1005

Hood L and Perlmutter R M(2004) The impact of systemsapproaches on biological problems



in drug discovery Nat Biotechnol22 1215ndash1217

Howes O D and Kapur S (2009)The dopamine hypothesis of schizo-phrenia version III ndash the final com-mon pathway Schizophr Bull 35549ndash562

Huang J T Leweke F M Oxley DWang L Harris N Koethe D et al(2006) Disease biomarkers in cere-brospinal fluid of patients with first-onset psychosis PLoS Med 3e428doi101371journalpmed0030428

Hwu H G Liu C M Fann C S Ou-Yang W C and Lee S F (2003)Linkage of schizophrenia with chro-mosome 1q loci in Taiwanese fami-lies Mol Psychiatry 8 445ndash452

Ideker T Galitski T and Hood L(2001) A new approach to decod-ing life systems biology Annu RevGenomics Hum Genet 2 343ndash372

Iwamoto K Kakiuchi C Bundo MIkeda K and Kato T (2004) Mole-cular characterization of bipolar dis-order by comparing gene expres-sion profiles of postmortem brainsof major mental disorders Mol Psy-chiatry 9 406ndash416

Jamshidi N and Palsson B O (2006)Systems biology of SNPs Mol SystBiol 2 38

Jiang C Xuan Z Zhao F andZhang M Q (2007) TRED atranscriptional regulatory elementdatabase new entries and otherdevelopment Nucleic Acids Res 35D137ndashD140

Jiang L Lindpaintner K Li H FGu N F Langen H He L etal (2003) Proteomic analysis ofthe cerebrospinal fluid of patientswith schizophrenia Amino Acids 2549ndash57

Jimenez-Marin A Collado-RomeroM Ramirez-Boo M Arce Cand Garrido J J (2009) Biolog-ical pathway analysis by ArrayUn-lock and Ingenuity Pathway Analy-sis BMC Proc 3(Suppl 4)S6doi1011861753-6561-3-S4-S6

Jones P and Cote R (2008) ThePRIDE proteomics identificationsdatabase data submission queryand dataset comparison MethodsMol Biol 484 287ndash303

Juhasz G Foldi I and PenkeB (2011) Systems biology ofAlzheimerrsquos disease how diversemolecular changes result in memoryimpairment in AD Neurochem Int58 739ndash750

Just M A Keller T A Malave V LKana R K and Varma S (2012)Autism as a neural systems disor-der a theory of frontal-posteriorunderconnectivity Neurosci Biobe-hav Rev 36 1292ndash1313

Kahraman A Avramov A NashevL G Popov D Ternes RPohlenz H D et al (2005) Phe-nomicDB a multi-species geno-typephenotype database for com-parative phenomics Bioinformatics21 418ndash420

Kana R K Libero L E and MooreM S (2011) Disrupted cortical con-nectivity theory as an explanatorymodel for autism spectrum disor-ders Phys Life Rev 8 410ndash437

Kanehisa M Goto S Sato Y Furu-michi M and Tanabe M (2012)KEGG for integration and interpre-tation of large-scale molecular datasets Nucleic Acids Res 40 D109ndashD114

Kang B Y Ko S and Kim D W(2010) SICAGO semi-supervisedcluster analysis using semanticdistance between gene pairs ingene ontology Bioinformatics 261384ndash1385

Karlsson R M Tanaka K Heilig Mand Holmes A (2008) Loss of glialglutamate and aspartate transporter(excitatory amino acid transporter1) causes locomotor hyperactivityand exaggerated responses to psy-chotomimetics rescue by haloperi-dol and metabotropic glutamate 23agonist Biol Psychiatry 64 810ndash814

Karlsson R M Tanaka K SaksidaL M Bussey T J Heilig Mand Holmes A (2009) Assessmentof glutamate transporter GLAST(EAAT1)-deficient mice for phe-notypes relevant to the negativeand executivecognitive symptomsof schizophrenia Neuropsychophar-macology 34 1578ndash1589

Karsenti E (2012) Towards an ldquooceanssystems biologyrdquo Mol Syst Biol 8575

Kerrien S Aranda B Breuza LBridgeA Broackes-Carter F ChenC et al (2012) The IntAct mole-cular interaction database in 2012Nucleic Acids Res 40 D841ndashD846

Khandaker G M Zimbron J LewisG and Jones P B (2012) Prenatalmaternal infection neurodevelop-ment and adult schizophrenia a sys-tematic review of population-basedstudies Psychol Med 1ndash19

King R Raese J D and BarchasJ D (1981) Catastrophe theoryof dopaminergic transmission arevised dopamine hypothesis ofschizophrenia J Theor Biol 92373ndash400

Kirov G Gumus D Chen W Nor-ton N Georgieva L Sari Met al (2008) Comparative genomehybridization suggests a role forNRXN1 and APBA2 in schizophre-nia Hum Mol Genet 17 458ndash465

Kirov G Pocklington A J HolmansP Ivanov D Ikeda M Ruderfer Det al (2012) De novo CNV analysisimplicates specific abnormalities ofpostsynaptic signalling complexes inthe pathogenesis of schizophreniaMol Psychiatry 17 142ndash153

KitabayashiY Narumoto J and FukuiK (2007) Neurodegenerative dis-orders in schizophrenia PsychiatryClin Neurosci 61 574ndash575

Kitano H (2002) Systems biologya brief overview Science 2951662ndash1664

Kobeissy F H Sadasivan S Liu JGold M S and Wang K K(2008) Psychiatric research psy-choproteomics degradomics andsystems biology Expert Rev Pro-teomics 5 293ndash314

Komek K Bard Ermentrout GWalker C P and Cho R Y (2012)Dopamine and gamma band syn-chrony in schizophrenia ndash insightsfrom computational and empiri-cal studies Eur J Neurosci 362146ndash2155

Korolainen M A Nyman T A Ait-tokallio T and Pirttila T (2010)An update on clinical proteomics inAlzheimerrsquos research J Neurochem112 1386ndash1414

Kotera M Yamanishi Y Moriya YKanehisa M and Goto S (2012)GENIES gene network inferenceengine based on supervised analysisNucleic Acids Res 40 W162ndashW167

Krull M Pistor S Voss N Kel AReuter I Kronenberg D et al(2006) TRANSPATH an informa-tion resource for storing and visu-alizing signaling pathways and theirpathological aberrations NucleicAcids Res 34 D546ndashD551

Lemer CAntezana E Couche F FaysF Santolaria X Janky R et al(2004) The aMAZE LightBench aweb interface to a relational databaseof cellular processes Nucleic AcidsRes 32 D443ndashD448

Le-Niculescu H Balaraman Y PatelS Tan J Sidhu K Jerome RE et al (2007) Towards under-standing the schizophrenia codean expanded convergent functionalgenomics approach Am J MedGenet B Neuropsychiatr Genet144B 129ndash158

Le-Niculescu H Patel S D Bhat MKuczenski R Faraone S V TsuangM T et al (2009) Convergentfunctional genomics of genome-wide association data for bipo-lar disorder comprehensive identi-fication of candidate genes path-ways and mechanisms Am J MedGenet B Neuropsychiatr Genet150B 155ndash181

Lescuyer P Hochstrasser D and Rabil-loud T (2007) How shall we usethe proteomics toolbox for bio-marker discovery J Proteome Res 63371ndash3376

Lett T A Tiwari A K MeltzerH Y Lieberman J A Potkin SG Voineskos A N et al (2011)The putative functional rs1045881marker of neurexin-1 in schizophre-nia and clozapine response Schizo-phr Res 132 121ndash124

Levin Y Wang L Schwarz E KoetheD Leweke F M and Bahn S(2010) Global proteomic profilingreveals altered proteomic signaturein schizophrenia serum Mol Psychi-atry 15 1088ndash1100

Linden D E (2012) The challengesand promise of neuroimaging inpsychiatry Neuron 73 8ndash22

Lopez A D and Murray C C (1998)The global burden of disease 1990-2020 Nat Med 4 1241ndash1243

Lucas M Laplaze L and Bennett MJ (2011) Plant systems biology net-work matters Plant Cell Environ 34535ndash553

Macaluso M and Preskorn S H(2012) How biomarkers will changepsychiatry from clinical trials topractice Part I introduction J Psy-chiatr Pract 18 118ndash121

Major Depressive Disorder Work-ing Group of the PsychiatricGWAS Consortium (2012) Amega-analysis of genome-wideassociation studies for majordepressive disorder Mol Psychiatrydoi 101038mp201221

Malaspina D (2006) Schizophrenia aneurodevelopmental or a neurode-generative disorder J Clin Psychia-try 67 e07

Manji H K and Lenox R H (2000)The nature of bipolar disorderJ Clin Psychiatry 61(Suppl 13)42ndash57

Martins-de-Souza D (2010) Proteomeand transcriptome analysis sug-gests oligodendrocyte dysfunction inschizophrenia J Psychiatr Res 44149ndash156

Martins-de-Souza D Alsaif M ErnstA Harris L W Aerts N LenaertsI et al (2012) The application ofselective reaction monitoring con-firms dysregulation of glycolysisin a preclinical model of schiz-ophrenia BMC Res Notes 5146doi1011861756-0500-5-146

Martins-de-Souza D Lebar M andTurck C W (2011) Proteomeanalyses of cultured astrocytestreated with MK-801 and clozapinesimilarities with schizophrenia EurArch Psychiatry Clin Neurosci 261217ndash228



Mazza T Ballarini P Guido R andPrandi D (2012) The relevanceof topology in parallel simulationof biological networks IEEEACMTrans Comput Biol Bioinform 9911ndash923

Mendes P Hoops S Sahle S GaugesR Dada J and Kummer U (2009)Computational modeling of bio-chemical networks using COPASIMethods Mol Biol 500 17ndash59

Mi H and Thomas P (2009) PAN-THER pathway an ontology-basedpathway database coupled with dataanalysis tools Methods Mol Biol563 123ndash140

Miller B J Culpepper N Rapa-port M H and Buckley P (2012)Prenatal inflammation and neu-rodevelopment in schizophrenia areview of human studies Prog Neu-ropsychopharmacol Biol PsychiatryPMID 22510462 [Epub ahead ofprint]

Mitchell K J (2011) The genetics ofneurodevelopmental disease CurrOpin Neurobiol 21 197ndash203

Moghaddam B and Javitt D (2012)From revolution to evolution theglutamate hypothesis of schizophre-nia and its implication for treatmentNeuropsychopharmacology 37 4ndash15

Moraru Ii Schaff J C Slepchenko BM and Loew L M (2002) Thevirtual cell an integrated modelingenvironment for experimental andcomputational cell biology Ann NY Acad Sci 971 595ndash596

Morganella S and Ceccarelli M(2012) VegaMC a Rbioconductorpackage for fast downstream analysisof large array comparative genomichybridization datasets Bioinformat-ics 28 2512ndash2514

Morris S E Holroyd C B Mann-Wrobel M C and Gold J M(2011) Dissociation of responseand feedback negativity in schiz-ophrenia electrophysiologicaland computational evidence fora deficit in the representation ofvalue Front Hum Neurosci 5123doi103389fnhum201100123

Munk C Sommer A F and Konig R(2011) Systems-biology approachesto discover anti-viral effectors ofthe human innate immune responseViruses 3 1112ndash1130

Nakatani N Hattori E Ohnishi TDean B IwayamaY Matsumoto Iet al (2006) Genome-wide expres-sion analysis detects eight genes withrobust alterations specific to bipo-lar I disorder relevance to neuronalnetwork perturbation Hum MolGenet 15 1949ndash1962

Nanavati D Austin D R Catapano LA Luckenbaugh D A Dosemeci

A Manji H K et al (2011) Theeffects of chronic treatment withmood stabilizers on the rat hip-pocampal post-synaptic density pro-teome J Neurochem 119 617ndash629

Nesvaderani M Matsumoto I andSivagnanasundaram S (2009)Anterior hippocampus in schizo-phrenia pathogenesis molecularevidence from a proteome studyAust N Z J Psychiatry 43 310ndash322

Neumann D Spezio M L Piven Jand Adolphs R (2006) Lookingyou in the mouth abnormal gaze inautism resulting from impaired top-down modulation of visual atten-tion Soc Cogn Affect Neurosci 1194ndash202

Nohesara S Ghadirivasfi MMostafavi S Eskandari M RAhmadkhaniha H ThiagalingamS et al (2011) DNA hypomethy-lation of MB-COMT promoter inthe DNA derived from saliva inschizophrenia and bipolar disorderJ Psychiatr Res 45 1432ndash1438

Noriega G (2007) Self-organizingmaps as a model of brain mecha-nisms potentially linked to autismIEEE Trans Neural Syst Rehabil Eng15 217ndash226

Noriega G (2008) A neural modelfor compensation of sensory abnor-malities in autism through feedbackfrom a measure of global percep-tion IEEE Trans Neural Netw 191402ndash1414

Ogden C A Rich M E Schork NJ Paulus M P Geyer M A LohrJ B et al (2004) Candidate genespathways and mechanisms for bipo-lar (manic-depressive) and relateddisorders an expanded convergentfunctional genomics approach MolPsychiatry 9 1007ndash1029

Olivier B G and Snoep J L(2004) Web-based kinetic model-ling using JWS online Bioinformat-ics 20 2143ndash2144

Ori A Wilkinson M C and Fer-nig D G (2011) A systems biol-ogy approach for the investiga-tion of the heparinheparan sul-fate interactome J Biol Chem 28619892ndash19904

OrsquoTuathaigh C M Desbonnet LMoran P M and WaddingtonJ L (2012) Susceptibility genesfor schizophrenia mutant mod-els endophenotypes and psychobi-ology Curr Top Behav Neurosci 12209ndash250

Pagel P Kovac S Oesterheld MBrauner B Dunger-Kaltenbach IFrishman G et al (2005) TheMIPS mammalian protein-proteininteraction database Bioinformatics21 832ndash834

Pennington K Dicker P Dunn M Jand Cotter D R (2008) Proteomicanalysis reveals protein changeswithin layer 2 of the insular cor-tex in schizophrenia Proteomics 85097ndash5107

Perez-Neri I Ramirez-Bermudez JMontes S and Rios C (2006) Pos-sible mechanisms of neurodegener-ation in schizophrenia NeurochemRes 31 1279ndash1294

Pollard M Varin C Hrupka B Pem-berton D J Steckler T and Sha-ban H (2012) Synaptic transmis-sion changes in fear memory cir-cuits underlie key features of an ani-mal model of schizophrenia BehavBrain Res 227 184ndash193

Ptak C and Petronis A (2010) Epige-netic approaches to psychiatric dis-orders Dialogues Clin Neurosci 1225ndash35

Qi Z Miller G W and Voit EO (2008) A mathematical modelof presynaptic dopamine homeosta-sis implications for schizophreniaPharmacopsychiatry 41(Suppl 1)S89ndashS98

Qi Z Miller G W and Voit E O(2010) Computational modeling ofsynaptic neurotransmission as a toolfor assessing dopamine hypothesesof schizophrenia Pharmacopsychia-try 43(Suppl 1) S50ndashS60

Rapoport J L Addington A M Fran-gou S and Psych M R (2005)The neurodevelopmental model ofschizophrenia update 2005 MolPsychiatry 10 434ndash449

Reacutenyi A (1959) On measures ofdependence Acta Math Hung 10441ndash451

Ripke S Sanders A R Kendler K SLevinson D F Sklar P HolmansP A et al (2011) Genome-wideassociation study identifies five newschizophrenia loci Nat Genet 43969ndash976

Robeva R (2010) Systems biology ndashold concepts new science newchallenges Front Psychiatry 11doi103389fpsyt201000001

Rotaru D C Yoshino H Lewis DA Ermentrout G B and Gonzalez-Burgos G (2011) Glutamate recep-tor subtypes mediating synapticactivation of prefrontal cortex neu-rons relevance for schizophrenia JNeurosci 31 142ndash156

Saetzler K Sonnenschein C andSoto A M (2011) Systems biol-ogy beyond networks generatingorder from disorder through self-organization Semin Cancer Biol 21165ndash174

Sales G Calura E Cavalieri D andRomualdi C (2012) Graphite ndasha Bioconductor package to convert

pathway topology to gene net-work BMC Bioinformatics 1320doi1011861471-2105-13-20

Salomonis N Hanspers K ZambonA C Vranizan K Lawlor S CDahlquist K D et al (2007) Gen-MAPP 2 new features and resourcesfor pathway analysis BMC Bioin-formatics 8217 doi1011861471-2105-8-217

Satta R Maloku E Zhubi A PibiriF Hajos M Costa E et al (2008)Nicotine decreases DNA methyl-transferase 1 expression and glu-tamic acid decarboxylase 67 pro-moter methylation in GABAergicinterneurons Proc Natl Acad SciUSA 105 16356ndash16361

Satuluri V Parthasarathy S and UcarD (2010) ldquoMarkov clustering ofprotein interaction networks withimproved balance and scalabilityrdquo inProceedings of the First ACM Interna-tional Conference on Bioinformaticsand Computational Biology (NiagaraFalls NY ACM) 247ndash256

Sayed Y and Garrison J M (1983)The dopamine hypothesis of schizo-phrenia and the antagonistic actionof neuroleptic drugs ndash a reviewPsychopharmacol Bull 19 283ndash288

Schork N J Greenwood T A andBraff D L (2007) Statistical genet-ics concepts and approaches inschizophrenia and related neuropsy-chiatric research Schizophr Bull 3395ndash104

Schwarz E Guest P C Steiner JBogerts B and Bahn S (2012)Identification of blood-based mol-ecular signatures for prediction ofresponse and relapse in schizophre-nia patients Transl Psychiatry 2e82

Seamans J K and Yang C R (2004)The principal features and mecha-nisms of dopamine modulation inthe prefrontal cortex Prog Neuro-biol 74 1ndash58

Seeman P (1987) Dopamine recep-tors and the dopamine hypothesis ofschizophrenia Synapse 1 133ndash152

Sen P K (2008) Kendallrsquos tau inhigh-dimensional genomic parsi-mony Inst Math Stat Collect 3251ndash266

Sequeira A Maureen M V andVawter M P (2012) The first decadeand beyond of transcriptional profil-ing in schizophrenia Neurobiol Dis45 23ndash36

Shannon P Markiel A Ozier OBaliga N S Wang J T Ram-age D et al (2003) Cytoscapea software environment for inte-grated models of biomolecular inter-action networks Genome Res 132498ndash2504



Simons N D Lorenz J G SheeranL K Li J H Xia D P andWagner R S (2012) Noninvasivesaliva collection for DNA analysesfrom free-ranging Tibetan macaques(Macaca thibetana) Am J Primatol74 1064ndash1070

Smith K (2011) Trillion-dollar braindrain Nature 478 15

Snyder S H (1976) The dopaminehypothesis of schizophrenia focuson the dopamine receptor Am JPsychiatry 133 197ndash202

Spencer K M (2009) The func-tional consequences of corticalcircuit abnormalities on gammaoscillations in schizophreniainsights from computational mod-eling Front Hum Neurosci 333doi103389neuro090332009

Stahl S M (2007) Beyond thedopamine hypothesis to the NMDAglutamate receptor hypofunctionhypothesis of schizophrenia CNSSpectr 12 265ndash268

Takahashi K Ishikawa N SadamotoY Sasamoto H Ohta SShiozawa A et al (2003) E-cell2 multi-platform E-cell simula-tion system Bioinformatics 191727ndash1729

Theocharidis A van Dongen SEnright A J and Freeman T C(2009) Network visualization andanalysis of gene expression datausing BioLayout Express(3D) NatProtoc 4 1535ndash1550

Thomas E A Dean B ScarrE Copolov D and Sutcliffe JG (2003) Differences in neu-roanatomical sites of apoD elevationdiscriminate between schizophreniaand bipolar disorder Mol Psychiatry8 167ndash175

Tretter F and Gebicke-Haerter P J(2012) Systems biology in psychi-atric research from complex datasets over wiring diagrams to com-puter simulations Methods MolBiol 829 567ndash592

Triesch J Teuscher C Deak G Oand Carlson E (2006) Gaze follow-ing why (not) learn it Dev Sci 9125ndash147

Ullah M Schmidt H Cho K Hand Wolkenhauer O (2006) Deter-ministic modelling and stochasticsimulation of biochemical pathwaysusing MATLAB Syst Biol (Steve-nage) 153 53ndash60

Van Hemert J L and Dicker-son J A (2010) PathwayAc-cess CellDesigner plugins for path-way databases Bioinformatics 262345ndash2346

Vattikuti S and Chow C C (2010)A computational model for cerebralcortical dysfunction in autism spec-trum disorders Biol Psychiatry 67672ndash678

Vawter M P Thatcher L Usen NHyde T M Kleinman J E andFreed W J (2002) Reduction ofsynapsin in the hippocampus ofpatients with bipolar disorder andschizophrenia Mol Psychiatry 7571ndash578

Vierling-Claassen D Siekmeier PStufflebeam S and Kopell N(2008) Modeling GABA alterationsin schizophrenia a link betweenimpaired inhibition and alteredgamma and beta range auditoryentrainment J Psychiatr Res 992656ndash2671

Villoslada P and Baranzini S (2012)Data integration and systemsbiology approaches for biomarker

discovery challenges and oppor-tunities for multiple sclerosis JNeuroimmunol 248 58ndash65

Volman V Behrens M M andSejnowski T J (2011) Downreg-ulation of parvalbumin at corti-cal GABA synapses reduces networkgamma oscillatory activity J Neu-rosci 31 18137ndash18148

von Eichborn J Bourne P E andPreissner R (2011) Cobweb aJava applet for network explorationand visualisation Bioinformatics 271725ndash1726

Waltz J A Frank M J WieckiT V and Gold J M (2011)Altered probabilistic learning andresponse biases in schizophreniabehavioral evidence and neurocom-putational modeling Neuropsychol-ogy 25 86ndash97

Westerhoff H V (2011) Systems biol-ogy left and right Meth Enzymol500 3ndash11

Westerhoff H V and Palsson BO (2004) The evolution ofmolecular biology into systemsbiology Nat Biotechnol 221249ndash1252

WHO (2009) The Global Burden of Dis-ease 2004 Update Geneva WorldHealth Organization

Woods A G Sokolowska I TaurinesR Gerlach M Dudley E ThomeJ et al (2012) Potential biomarkersin psychiatry focus on the choles-terol system J Cell Mol Med 161184ndash1195

Zhang J Goodlett D R andMontine T J (2005) Pro-teomic biomarker discovery incerebrospinal fluid for neurodegen-erative diseases J Alzheimers Dis 8377ndash386

Zhang Z Larner S F KobeissyF Hayes R L and Wang KK (2010) Systems biology andtheranostic approach to drug dis-covery and development to treattraumatic brain injury Methods MolBiol 662 317ndash329

Zhu X Gerstein M and Sny-der M (2007) Getting connectedanalysis and principles of biologicalnetworks Genes Dev 21 1010ndash1024

Conflict of Interest Statement Theauthors declare that the research wasconducted in the absence of any com-mercial or financial relationships thatcould be construed as a potential con-flict of interest

Received 26 August 2012 paper pendingpublished 10 September 2012 accepted06 December 2012 published online 24December 2012Citation Alawieh A Zaraket FA Li J-L Mondello S Nokkari A RazafshaM Fadlallah B Boustany R-M andKobeissy FH (2012) Systems biologybioinformatics and biomarkers in neu-ropsychiatry Front Neurosci 6187 doi103389fnins201200187This article was submitted to Frontiers inSystems Biology a specialty of Frontiersin NeuroscienceCopyright copy 2012 Alawieh Zaraket LiMondello Nokkari Razafsha FadlallahBoustany and Kobeissy This is an open-access article distributed under the termsof the Creative Commons AttributionLicense which permits use distributionand reproduction in other forums pro-vided the original authors and sourceare credited and subject to any copy-right notices concerning any third-partygraphics etc



several multifactorial environmental and genetic factors (Schorket al 2007 Cacabelos et al 2011 Gormanns et al 2011 Mitchell2011 Ripke et al 2011) Unlike classical Mendelian inherited dis-easesa single gene mutation cannot explain the overt phenotype ofNP disorders Clinical appearances generally result from the con-current influence of different genes along with several epigeneticmechanisms (Cowan et al 2002 Behan et al 2009 Cacabeloset al 2011)

Many other limitations circumscribe the current study of NPdisorders including the inaccessibility to the brain tissue (Levinet al 2010 Villoslada and Baranzini 2012) the overlap of symp-toms among conditions that can result in diagnostic ambiguityand the inability to identify specific diagnostic biomarkers usingcurrent imaging techniques like functional Magnetic ResonanceImaging (fMRI) Positron Emission Tomography (PET-scan) Sin-gle Photon Emission Computed Tomography (SPECT) and others(refer to the review by Linden 2012) Moreover current avail-able diagnostic systems like DSM-IV and ICP-10 rely principallyon cliniciansrsquo assessment even if different disorders ndash treated dif-ferently ndash have similar manifestations (Linden 2012 Tretter andGebicke-Haerter 2012) Molecular biology techniques may enablethese diagnostic strategies to be more accurate and objective andallow for disease classifications better suited for more targetedpersonalized and effective therapy

With these aims in mind the previous molecular biologyapproach that investigates the effect of a single gene or proteinappears insufficient to establish a complete understanding of theNP disorders (Kitano2002 Zhang et al 2010) Such a reductionistapproach fails to provide an insight into the structure and dynam-ics of biological systems Biological systems are characterized byhierarchy (having several layers of organization with interactionsamong and between layers) emergence (having properties thatare visible at the system level and unexpected from the underly-ing components alone) and robustness (having the capacity ofthe system or network to operate normally under a wide range ofconditions) Meanwhile the reductionist approach lags behind theability to detect emergent properties and combinatorial effects thatcharacterize biological systems It also falls short of understand-ing compensatory and synergistic responses of the system towardperturbations (Lucas et al 2011 Munk et al 2011 Saetzler et al2011 Westerhoff 2011) Systems biology would address the mul-tifactorial aspect of NP diseases to better chart the functionalityof complex biological systems for insights into the etiology andpathophysiology

In this review we will be discussing the potentials of systemsbiology bioinformatics and biomarker research in the area of NPdisorders Furthermore we will be outlining how the knowledgegained through these methodologies can be translated into clin-ical use providing clinicians with improved ability to diagnosemanage and treat NP patients

THE APPROACH OF SYSTEMS BIOLOGYThe revolution of systems biology in the twentieth century rep-resented a ldquoparadigm shiftrdquo of molecular biology from a reduc-tionist approach to a holistic approach (Westerhoff and Palsson2004) The new approach tries to analyze the relationships withina biological system to find how different components of that

particular system interact (Kitano 2002 Hood and Perlmutter2004 Fang and Casadevall 2011 Lucas et al 2011 Westerhoff2011 Karsenti 2012) This accelerating ldquoshiftrdquo was grounded ontwo lines of thought whose integration is believed to have trans-formed molecular biology into the more sophisticated disciplineof systems biology (Westerhoff and Palsson 2004) These are thesequencing of the human genome and the great advances in thehigh-throughput (HT) discovery tools that allowed collection andanalysis of large data sets of genomics proteomics and others(Ideker et al 2001 Kitano 2002 Food and Drug Administra-tion 2011 Ori et al 2011) It is believed that systems biology willhelp understand and simplify the complexity of biological systemsand the available datasets obtained by HT techniques Such com-plexity is impeding the development of new diagnostic tools andtherapeutics for several complex diseases especially in the areas ofneuropsychiatry (Hood and Perlmutter 2004 Robeva 2010 Fangand Casadevall 2011 Westerhoff 2011)

Systems biology works by combining mathematical modelswith experimental molecular information from in silico in vivoand in vitro studies with HT data sets including genomicsproteomics metabolomics and transcriptomics (Robeva 2010Zhang et al 2010 Westerhoff 2011) For this purpose the integra-tive use of computational tools bioinformatics and engineeringsystems analysis represents the working tools in systems biologyTherefore systems biology requires a strong computational infra-structure and simulation software tools (Kitano 2002 Hood andPerlmutter 2004) that are able to handle huge databases and deter-mine dependencies that can be correlated with biological functions(Jamshidi and Palsson 2006) In addition HT genomics pro-teomics and metabolomics infrastructure are needed to achieverobustness and reproducibility (Kitano 2002 Hood and Perl-mutter 2004 Westerhoff and Palsson 2004 Jamshidi and Pals-son 2006) Finally enough experimental data should be gath-ered to provide raw material for analysis and to validate presentresults generated from multidisciplinary fields of mathematicsengineering bioinformatics and medicine (Robeva 2010)

The holistic analysis of systems biology aims to solve the previ-ously mentioned limitations of the reductionist approach Theseinclude (1) the inadequacy to investigate the hierarchy robust-ness and emergence characteristics of a biological system (2)the incompetence in understanding and reconstituting systemrsquosdynamics and (3) the failure to decipher the complexity of themassive data output from HT techniques (Kitano 2002 Zhanget al 2010 Lucas et al 2011) Therefore systems biology pro-vides a mean to understand the normal functioning of the systemand to predict the systemsrsquo response to perturbations (Idekeret al 2001 Kobeissy et al 2008) With this approach systemsbiology increases diagnostic prognostic and disease-monitoringpotentials for clinical applications (Hood and Perlmutter 2004)

The holistic approach of systems biology can be either a top-bottom approach starting from ldquoomicsrdquo datasets and drawinginferences related to the flow of information within biological net-works or bottom-top approach starting from experimental mole-cular data to draw models of these networks (Jamshidi and Palsson2006 Fang and Casadevall 2011 Lucas et al 2011 Wester-hoff 2011) The networks and sub-networks that systems biologyaims to study represent modules in the biological systems These

































1 m(0)2 m(0)



A(t )i =

xS




(1)


m(t+1)i =

1∣∣∣C (t )i

∣∣∣sum

xjisin C (t )i

C (t )i xj (2)















n (n minus 1)(3)



















































































































































































































































































































1 m(0)2 m(0)



A(t )i =

xS




(1)


m(t+1)i =

1∣∣∣C (t )i

∣∣∣sum

xjisin C (t )i

C (t )i xj (2)















n (n minus 1)(3)






























































































































































































































































































n (n minus 1)(3)




















































































































































































































































































Systems Biology, Bioinformatics, and Biomarkers in Neuropsychiatry

Documents

Transcript of Systems Biology, Bioinformatics, and Biomarkers in Neuropsychiatry