review article translational bioinformatics for diagnostic

Hindawi Publishing CorporationBioMed Research InternationalVolume 2013 Article ID 901578 13 pageshttpdxdoiorg1011552013901578

Review ArticleTranslational Bioinformatics for Diagnostic andPrognostic Prediction of Prostate Cancer in the Next-GenerationSequencing Era

Jiajia Chen12 Daqing Zhang1 Wenying Yan1 Dongrong Yang3 and Bairong Shen1

1 Center for Systems Biology Soochow University Suzhou 215006 China2 School of Chemistry Biology and Material Engineering Suzhou University of Science and Technology Suzhou 215011 China3Department of Urology The Second Affiliated Hospital of Soochow University Suzhou 215004 China

Correspondence should be addressed to Bairong Shen bairongshensudaeducn

Received 1 May 2013 Accepted 22 June 2013

Academic Editor Xinghua Lu

Copyright copy 2013 Jiajia Chen et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

The discovery of prostate cancer biomarkers has been boosted by the advent of next-generation sequencing (NGS) technologiesNevertheless many challenges still exist in exploiting the flood of sequence data and translating them into routine diagnostics andprognosis of prostate cancer Here we review the recent developments in prostate cancer biomarkers by high throughput sequencingtechnologies We highlight some fundamental issues of translational bioinformatics and the potential use of cloud computing inNGS data processing for the improvement of prostate cancer treatment

1 Introduction

Prostate cancer (PCa) is the most common cancer andthe second leading cause of cancer deaths among males inwestern societies [1] It is estimated that 241740 newPCa caseswere diagnosed and that 28170men died from it in theUnitedStates in 2012 Since its discovery over 20 years ago ProstateSpecific Antigen (PSA) has been the mainstay for diagnosisand prognosis of prostate cancer However the routine useof PSA screening remains controversial owing to its limitedspecificity PSA fails to differentiate PCa from commonprostate disorders moreover it cannot discriminate betweenaggressive tumors and low-risk ones that may otherwisenever have been diagnosed without screening [2] As suchoverdetection and overtreatment represent critical conse-quences of PSA-based screening [3] The ongoing debatehighlights the need for more sensitive and specific tools toenable more accurate diagnosis and prognosis

During the last decade the ability to interrogate prostatecancer genomes has rapidly advanced The resolution forgenomic mutation discovery was improved first with array-based methods and now with next-generation sequencing(NGS) technologies These high throughput technologies

open up the possibility to individualize the diagnosis andtreatment of cancer However significant challenges partic-ularly with respect to integration storage and computationof large-scale sequencing data will have to be overcome totranslate NGS achievements into the bedside of the cancerpatient Translational informatics evolves as a promisingmethodology that can provide a foundation for crossing suchldquotranslational barriersrdquo [4 5]

Here we overview the NGS-based strategies in prostatecancer research with focus on upcoming biomarker can-didates that show promise for the diagnosis and prognosisof prostate cancer We also outline future perspectives fortranslational informatics and cloud computation to improveprostate cancer management

2 Microarray Based Diagnosis andPrognosis of PCa

In the past two decades high-throughput microarray profil-ing has been utilized to track complex molecular aberrationsduring PCa carcinogenesis We performed a comprehensivesearch in the Gene Expression Omnibus (GEO) for the

2 BioMed Research International

Table 1 Number of PCa-associated GEO series generated by microarray and NGS

Methodology Gene expressionprofiling

Noncoding RNAprofiling

Genomebindingoccupancy

profiling

Genome methylationprofiling

Genome variationprofiling

Microarray 266 34 17 21 35NGS 11 1 18 2 2

array-based profiles in human PCaThe retrieved GEO seriesgenerally fall into 5 categories gene expression profilingnoncoding RNA profiling genome bindingoccupancy pro-filing genome methylation profiling and genome variationprofiling The number of GEO series for each category issummarized in Table 1

Together these array-based technologies have shed lighton the genetic alterations in the PCa genome Among theabnormalities affecting prostate tumors the copy numberalteration is the most common one [6]

Numerous early studies have used comparative genomehybridization (CGH) or single nucleotide polymorphism(SNP) arrays to assess copy number changes in tumor DNAAs a result multiple genomic regions that displayed frequentgain or loss in the PCa genome [6ndash11] have been revealedChromosome 3p14 8p22 10q23 13q13 and 13q14 are foundto display broad copy number deletion Key genes mappingwithin these deleted regions include NKX31 PTEN [12]BRCA2 C13ORF15 SIAH3 [11] RB1 HSD17B2 [9] FOXP1RYBP and SHQ1 [6] High-level copy number gains aredetected at 5p13 14q21 7q22 Xq12 and 8q13 [9] Keyamplified genes mapping within these regions include SKP2FOXA1 AR [11] and HSD17B3 [9]

Usingmicroarray substantial efforts have also beenmadeto characterize prostate cancer gene expression profilesDifferentially expressed genes identified in these studies pointto a plethora of candidate biomarkers with diagnostic orprognostic value

Adiagnosticmarker is able to differentiate prostate cancerwith other prostatic abnormalities There are many emergingmarkers that show promise for PCa diagnosis such as alpha-methylacyl-CoA racemase (AMACR) [13] prostate cancergene 3 (PCA3) [14] early prostate cancer antigen (EPCA)-2[15] Hepsin [16] kallikrein-related peptidase 2 (KLK2) [17]and polycomb group protein enhancer of zeste homolog 2(EZH2) [18] The most prominent of these is PCA3 whichwas found to exhibit higher sensitivity and specificity forPCa detection than PSA PCA3 thus provides a potentialcomplement to PSA for the early diagnosis of PCa

Prognostic biomarkers for prediction of prostate cancerpatient outcome have also been identified Increased serumlevels of IL-6 and its receptor (IL-6R) are associated withmetastatic and hormone refractory disease [19] Elevatedserum chromogranin A levels are indicative of poor progno-sis and decreased survival [20] Other differentially expressedmolecules with prognostic potential include the urokinaseplasminogen activation (uPA) [21] TGF-1205731 [22 23] MUC1[24] CD24 [25] hCAP-D3 [26] vesicular monoamine trans-porter 2 (SLC18A2) [27] TEA domain family member 1

(TEAD1) c-Cbl [28] SOX7 and SOX9 [29] nuclear receptorbinding protein 1 (NRBP1) [30] CD147 [31] and Wnt5a [32]Each of thesemarkers will require proper validation to ensuretheir clinical utility

While microarray technology represents a wonderfulopportunity for the detection of genomic alterations there aresignificant issues thatmust be considered For example array-based methods are impossible to detect variations at a lowfrequency (many well below 1) in the samples In additionmicroarrays can only provide information about the genesthat are already included on the array The emerging next-generation sequencing technology also called massively par-allel sequencing however helps to overcome the challengesby generating actual sequence reads [33]

3 NGS Based Diagnosis and Prognosis of PCa

Key feature of the next-generation sequencing technologyis the massive parallelization of the sequencing process Byvirtue of the massively parallel process NGS generates hun-dreds of millions of short DNA reads (100ndash250 nucleotides)simultaneously which are then assembled and aligned toreference genomes

A number of NGS systems are available commerciallyincluding Genome AnalyzerHiSeq 2000MiSeq from Illu-mina SOLiDPGMProton from Life Sciences GS-FLX(454)GS Junior from Roche as well as novel single moleculesequencers for example Heliscope from Helicos Biosciencesand SMRT offered by Pacific Biosciences All these technolo-gies provide digital information onDNA sequences andmakeit feasible to discover genetic mutations at unprecedentedresolution and lower cost

It is generally accepted that cancers are caused by theaccumulation of genomic alterations NGS methods can wellsqueeze all the alteration information that remains hiddenwithin the genome including point mutations small inser-tions and deletions (InDels) copy number alterations (CNV)chromosomal rearrangements and epigenetic alterationsFor these reasons NGS has become the method of choice forlarge-scale detection of somatic cancer genome alterationsand is changing the way how cancer genome is analyzed AnNGS-based research pipeline for PCa biomarkers is given inFigure 1

Currently next-generation sequencing is being applied tocancer genome study in various ways

(1) genome-based sequencing (DNA-Seq) yielding in-formation on sequence variation InDels chromoso-mal rearrangements and copy number variations

BioMed Research International 3

Next-generation sequencing technologies

DNA-Seq Chip-Seq RNA-Seq Methyl-Seq

CNV Chromosomal rearrangement

Point mutation Alternative splicingGene fusion

TFBS

DNA methylationHistone modification

RNA expression and discovery

Pathway databases

Statistical methods

Regulatory networkCancer-related pathways

Molecularalteration

Literatures

i d di1

1

2

2

3 4 5 6 7 8 9 10 11 12 13 14

G U A

mRNA

5998400

CG

GC

C

GGCMM

MM

Zfp198 Elys

Err2

Rif1

BAF155

Wdr18Btbd14a

Arid3aTif1B

Sall1

Pelo

Zfp609

Mybbp

NF45SP1

HDAC2

Cdk1

EWS

Zfp219

Rybp

RequiemArid3b

Wapl

P66B

Rex1

Oct4 Zfp281

Nanog Dax1

Nac1

Sall4

REST

Rnf2

Rai14

Prmt1 YY1

MAPK

Figure 1 NGS-based pipeline for cancer marker discovery

(2) transcriptome-based sequencing (RNA-Seq) yield-ing quantitative information on transcribed regions(total RNA mRNA or noncoding RNAs)

(3) interactome-based sequencing (ChIP-Seq) yieldinginformation on protein binding sequences and his-tone modification

(4) methylome-based sequencing (Methl-seq) yieldingquantitative information on DNA methylation andchromatin conformation

In the following part we will introduce the main applica-tions to next-generation sequencing of prostate cancer usingexamples from the recent scientific literature (summarized inTable 2)

31 DNA-Seq According to the proportion of the genometargeted DNA-Seq is categorized into whole-genome seq-uencing and exome sequencing The goal of whole-genomesequencing is to sequence the entire genome not just cod-ing genes at a single-base resolution By whole-genomesequencing recent studies have provided detailed landscapeof genomic alterations in localized prostate cancers [6 1146 68 69] The full range of genomic alterations that driveprostate cancer development and progression including copynumber gains and losses single nucleotide substitutions andchromosomal rearrangements are readily identified

311 Copy Number Alteration Most prostate cancers exhibitsomatic copy number alterations with genomic deletions

outnumbering amplifications [6] Early methods for copynumber analysis involve fluorescence in situ hybridizationsand array-based methods (CGH arrays and SNP arrays)More recently NGS technologies have been utilized and offersubstantial benefits for copy number analysis

NGS used changes in sequencing depth (relative to anormal control) to identify copy number changesThe digitalnature of NGS therefore allows accurate estimation of copynumber levels at higher resolution In addition NGS canprovide novel gene copy information such as homozygousand heterozygous deletions and gene amplifications whereastraditional sequencing approaches cannot For exampleby next-generation sequencing of castrate-resistant prostatecancer (CRPC) Collins et al [34] identified a homozygous9p21 deletion spanning the MTAP CDKN2 and ARF genesand deficiency of MTAP was suggested as an exploitabletumor target

312 Somatic Nucleotide Substitutions While whole-genomesequencing provides the most comprehensive characteriza-tion of the cancer genome it is the most costly Alternativelytargeted sequencing approaches such as exome sequencingassemble multiple cancer genomes for analysis in a cost-effective manner Whole-exome sequencing captures thecoding exons of genes that contain the vastmajority of diseasecausing mutations Relative to structural alterations pointmutations are less common in prostate cancer [6 70] and theaverage mutation rate was estimated at 14Mbminus1 in localizedPCa [35] and 20Mbminus1 in CRPC [36]


Table 2 Summary of NGS-based studies on prostate cancer

Discoveries Method ReferencesCopy number loss of MTAP CDKN2 and ARF genes

DNA-Seq

[34]Somatic mutations in MTOR BRCA2 ARHGEF12 and CHD5 genes [11]NCOA2 p300 the AR corepressor NRIP1RIP140 and NCOR2SMRT [6]Somatic mutations in SPOP FOXA1 and MED12 [35]Somatic mutations in MLL2 and FOXA1 [36]Somatic mutations in TP53 DLK2 GPC6 and SDF4 [37]TMPRSS2ERG TMPRSS2ETV1

RNA-Seq

[38]TMPRSS2ETV4 [39]TMPRSS2ETV5 SLC45A3ETV5 [40]TMPRSS2ELK4 [41]SLC45A3ETV1 HERV-K 22q1123ETV1 HNRPA2B1ETV1 and C15ORF21ETV1 [42]KLK2ETV4 and CANT1ETV4 [43]SLC45A3BRAF or ESRP1RAF1 [44]C15orf21Myc [45]EPB41BRAF [46]TMEM79SMG5 [47]Differential expression of PCAT-1 [48]Differential expression of miR-16 miR-34a miR-126lowast miR-145 and miR-205 [49]HDACs and EZH2 work as ERG corepressors

Chip-Seq

[50]AP4 as a novel co-TF of AR [51]POU2F1 and NKX3-1 [52]Runx2a regulates secretion invasiveness and membrane secretion [53]A novel transcriptional regulatory network between NKX3-1 AR and the RAB GTPase signaling pathway [54]Distinct patterns of promoter methylation around transcription start sites Methyl-Seq [55]

Capillary-based exome sequencing has extensively beenperformed in localized PCa and CRPC and a handful ofoncogenic point mutations have been defined RemarkablyTaylor et al [6] performed focused exon resequencing in 218prostate cancer tumors and identified multiple somatic alter-ations in the androgen receptor (AR) gene as well as itsupstream regulators and downstream targets For examplethe AR coactivator NCOA2 and p300 the AR corepressorNRIP1RIP140 and NCOR2SMRT were found to harborsomatic mutations Other genes including KLF6 TP53 AREPHB2 CHEK2 and ATBF1 [6 71ndash74] have also been re-ported to harbor somaticmutations in localized prostate can-cer

RecentlyNGS is becoming increasingly routine for exomesequencing analysis Whole-exome sequencing using next-generation sequencing (NGS) technologies compares all exonsequences between tumors and matched normal samplesMultiple reads that show nonreference sequence are detectedas point mutations In this way a number of driver mutationsin prostate cancer have been uncovered Robbins et al [11]used NGS-based exome sequencing in 8 metastatic prostatetumors and revealed novel somatic point mutations in genesincluding MTOR BRCA2 ARHGEF12 and CHD5 Kumaret al [37] performed whole-exome sequencing of lethalmetastatic tumors and high-grade primary carcinomasTheyalso observed somatic mutations in TP53 DLK2 GPC6 andSDF4 More recently Barbieri et al [35] and Grasso et al [36]systematically analyzed somatic mutations in large cohorts of

prostate tumors Barbieri et al [35] investigated 112 primarytumor-normal pairs and revealed novel recurrent mutationsin SPOP FOXA1 and MED12 Grasso et al [36] sequencedthe exomes of 11 treatment-naive and 50 lethal CRPC andidentified recurrent mutations in multiple chromatin- andhistone-modifying genes includingMLL2 and FOXA1Thesetwo studies also reported mutated genes (SPOP [35] andCHD1 [36]) that may define prostate cancer subtypes whichare ETS gene family fusion negative

Together these findings present a comprehensive list ofspecific genes that might be involved in prostate cancer andprioritize candidates for future study

32 RNA-Seq In addition to genome applications NGS willalso dramatically enhance our ability to analyze transcrip-tomes Before NGS microarrays have been the dominanttechnology for transcriptome analysis Microarray technolo-gies rely on sequence-specific probe hybridization and flu-orescence detection to measure gene expression levels It issubject to high noise levels cross-hybridization and limiteddynamic range Compared to microarrays the emergingRNA-Seq provides digital gene expressionmeasurements thatoffer significant advantages in resolution dynamic range andreproducibility The goal of transcriptome sequencing is tosequence all transcribed genes including both coding andnoncoding RNAs It is independent of prior knowledge andoffers capacity to identify novel transcripts and mutations


that microarrays could not achieve such as fusion genesnoncoding RNAs and splice variants

321 Gene Fusions Recurrent gene fusion is a prevalenttype of mutation resulting from the chromosomal rearrange-ments which can generate novel functional transcripts thatserve as therapeutic targets Early studies relied on cytoge-netic methods to detect chromosomal rearrangements How-ever thismethod is only applicable in cases of simple genomesand is vulnerable in complex genomes of epithelial cancerssuch as PCa

Complete sequencing of prostate cancer genomes hasprovided further insight into chromosomal rearrangementsin prostate cancer NGS technologies for example paired-end sequencing approaches are sufficiently sensitive to detectbreak point crossing reads and are extremely powerful for thediscovery of fusion transcripts and potential break points

The first major recurrent fusion to be identified inprostate cancer was discovered by Tomlins et al using CancerOutlier Profile Analysis (COPA) algorithm [38] The fusiondiscovered places two oncogenic transcription factors fromthe ETS family (ETV1 and ERG) under control of theprostate-specific gene TMPRSS2

While the TMPRSS2ETV1 fusion is rare and occurs in1ndash10 of prostate cancers [75] the TMPRSS2ERG fusion ispresent in roughly half of prostate cancers and is the mostcommon genetic aberration so far described in prostatecancer Furthermore TMPRSS2ERG is unique only to pro-static nonbenign cancers [76] Given this high specificity itsclinical application as ancillary diagnostic test or prognosticbiomarker is promising The expression of TMPRSS2ERGfusion gene has been proposed as a diagnostic tool alone orin combination with PCA3 [77] In addition many studieshave suggested that TMRSS2ERG could be a prognosticbiomarker for aggressive prostate cancer

Following Tomlinsrsquo pioneering discovery subsequentresearch has identified a host of similar ETS family genefusions Other oncogenic ETS transcription factors forexample ETV4 [39] ETV5 [40] and ELK4 [41] have beenidentified as additional fusion partners for TMPRSS2 Otherunique 51015840 fusion partner genes to ETS family members havealso been identified such as SLC45A3 HERV-K 22q1123HNRPA2B1 and C15ORF21 in fusion with ETV1 [42] KLK2and CANT1 in fusion with ETV4 [43] and SLC45A3ETV5[40]

For ETS fusion-negative prostate cancers subtypesnovel gene fusions have also been identified includingSLC45A3BRAF ESRP1RAF1 SLC45A3BRAF ESRP1RAF1[44] C15orf21Myc [45] EPB41BRAF [46] andTMEM79SMG5 [47]

322 Noncoding RNAs In addition to gene fusions RNA-Seq also enables discovery of new noncoding RNAs (ncR-NAs)with the potential to serve as cancermarkers Transcrip-tome sequencing of a prostate cancer cohort has identified anunannotated ncRNA PCAT-1 as a transcriptional repressorlinked to PCa progression [48] RNA- Seq was also appliedto identify differentially expressed microRNAs (eg miR-16miR-34a miR-126lowast miR-145 and miR-205) associated with

metastatic prostate cancer [49] These findings establish theutility of RNA-Seq to identify disease-associated ncRNAs thatcould provide potential biomarkers or therapeutic targets

33 ChIP-Seq Another application of NGS capitalizes on theability to analyze protein-DNA interactions as for ChIP-SeqChIP-Seq provides clear indications of transcription factorbinding sites (TFBSs) at high resolution It is also well suitedfor detecting patterns of modified histones in a genome-widemanner [78]

Much of gene regulation occurs at the level of transcrip-tional control In addition aberrant histone modifications(methylation or acetylation) are also associated with cancerTherefore experimental identification of TFBSs or histonemodifications has been an area of high interest In traditionalChIP-chip approaches DNA associated with a transcriptionfactor or histone modification of interest is first selectivelyenriched by chromatin immunoprecipitation followed byprobing on DNA microarrays In contrast to ChIP-chipChIP-Seq uses NGS instead of custom-designed arrays toidentify precipitated DNA fragments thus yielding moreunbiased and sensitive information about target regions

Many efforts have employed ChIP-Seq approaches tocharacterize transcriptional occupancy of AR and manynovel translational partners of AR have been identified UsingChIP-Seq Chng et al [50] performed global analysis of ARand ERG binding sites They revealed that ERG promotesprostate cancer progression by working together with tran-scriptional corepressors including HDACs and EZH2 Zhanget al [51] developed a comotif scanning program calledCENTDIST and applied it on an AR ChIP-Seq dataset froma prostate cancer cell lineThey correctly predicted all knownco-TFs of AR as well as discovered AP4 as a novel AR co-TF Little et al [53] used genome-wide ChIP-Seq to studyRunx2 occupancy in prostate cancer cells They suggestednovel role of Runx2a in regulating secretion invasiveness andmembrane secretion Tan et al [54] showed that NKX3-1colocalizes with AR and proposed a critical transcriptionalregulatory network between NKX3-1 AR and the RABGTPase signaling pathway in prostate cancer Urbanucci etal [79] identified AR-binding sites and demonstrated thatthe overexpression of AR enhances the receptor binding tochromatin in CRPC By comparing nucleosome occupancymaps using nucleosome-resolution H3K4me2 ChIP-Seq Heet al [52] found that nucleosome occupancy changes canpredict transcription factor cistromes This approach alsocorrectly predicted the binding of two factors POU2F1 andNKX3-1 By high-resolutionmapping of intra- and interchro-mosome interactions Rickman et al [80] demonstrated thatERG binding is enriched in hotspots of differential chromatininteractionTheir result indicated that ERG overexpression iscapable of inducing changes in chromatin structures

Taken together these studies have provided a completeregulatory landscape in prostate cancer

34Methyl-Seq Another fertile area forNGS involves assess-ment of the genome-widemethylation status of DNAMethy-lation of cytosine residues in DNA is known to silenceparts of the genome by inducing chromatin condensation


DNA hypermethylation probably remains most stable andabundant epigenetic marker

Several DNA methylation markers have been identifiedin prostate cancer The most extensively studied one isCpG island hypermethylation of glutathione-S-transferaseP (GSTP1) promoter DNA resulting in the loss of GSTP1expression [81] Today GSTP1 hypermethylation is most fre-quently evaluated as diagnostic biomarker for prostate cancerIt is also an adverse prognosticmarker that predicts relapse ofpatients following radical prostatectomy [82]

With the aid ofNGS technologies genome-widemappingof methylated cytosine patterns in cancer cells becomefeasible Based on established epigenetics methods there areemerging next-generation sequencing applications for theinterrogation of methylation patterns including methyl-ation-dependent immunoprecipitation sequencing (MeDIP-Seq) [83] cytosine methylome sequencing (MethylC-Seq)[84] reduced representation bisulfite sequencing (RRBS-Seq)[85] methyl-binding protein sequencing (MBP-Seq) [86]and methylation sequencing (Methyl-Seq) [87] A number ofaberrant methylation profiles have been developed so far andare being evaluated as potential markers for early diagnosisand risk assessment As an example using MethylPlex-SeqKim et al [55] mapped the global DNAmethylation patternsin prostate tissues and revealed distinct patterns of promotermethylation around transcription start sites The compre-hensive methylome map will further our understanding ofepigenetic regulation in prostate cancer progression

4 PCa Biomarkers in Combinations

Diagnostic and prognostic markers with relevance to PCaare routinely identified Nevertheless we should stay awarethat candidate markers obtained are often irreproduciblefrom experiment to experiment and very few molecules willmake it to the routine clinical practice Cancer is a nonlineardynamic system that involves the interaction of many bio-logical components and is not driven by individual causativemutations In most cases no single biomarker is likely to dic-tate diagnosis or prognosis success Consequently the futureof cancer diagnosis and prognosis might rely on the combi-nation of a panel of markers Kattan and associates [88] haveestablished a prognosticmodel that incorporates serumTGF-1205731 and IL-6R for prediction of recurrence following radicalprostatectomy This combination shows increased predictiveaccuracy from 75 to 84 Furthermore a predictive modelincorporating GSTP1 retinoic acid receptor 1205732 (RAR 1205732)and adenomatous polyposis coli (APC) has been assessed Butno increased diagnostic accuracy was shown compared withPSA level alone [89]More recently a panel ofmarkers PCA3serum PSA level and free PSA show improved predictivevalue compared with PSA level and free PSA Again theclinical utility of these combinations needs to be evaluated inlarge-scale studies

5 PCa Biomarkers at Pathway Level

Recently an importance of pathway analysis has been empha-sized in the study of cancer biomarkers Pathway-basedapproach allows biologists to detect modest expression

changes of functionally important genes thatwould bemissedin expression-alone analysis In addition this approachenables the incorporation of previously acquired biologicalknowledge and makes a more biology-driven analysis of ge-nomics data Pathway analysis typically correlates a given setof molecular changes (eg differential expression mutationand copy number variation data) by projecting them ontowell-characterized biological pathways A number of curateddatabases are available for canonical signaling and metabolicpathways such as Kyoto Encyclopedia of Genes and Genom-es (KEGG httpwwwgenomejpkegg) Molecular Signa-tures Database (MSigDB) IngenuityPathway Analysis (IPAhttpwwwingenuitycom) GeneGO by MetaCore (httpwwwgenegocom) and Gene Set Enrichment Analysis(GSEA httpwwwbroadinstituteorggsea) Enrichment ofpathways can be evaluated by overrepresentation statis-tics The overall flowchart of the proposed pathway-basedbiomarker approach is illustrated in Figure 1 Using thispipeline the pathways enriched with aberrations are identi-fied and then proposed to be the potential candidatemarkers

Some impressive progress has been made to identifypathways with relevance to the pathophysiology of prostatecancer Rhodes et al [90] were of the first to perform pathwayanalysis of the microarray expression datasets By meta-analysis of 4 independentmicroarray datasets they generateda cohort of genes that were commonly deregulated in PCaThe authors then mapped the identified deregulated genes tofunctional annotations and pinpointed polyamine and purinebiosynthesis as critical pathway altered in PCa Activationof Wnt signaling pathway was reported to be key pathwaysdefining the poor PCa outcome group [91] A comparisonof castration-resistant and castration-sensitive pairs of tumorlines highlighted theWnt pathway as potentially contributingto castration resistance [37] Using a similar pathway-basedapproachWang et al found Endothelin-1EDNRA transacti-vation of the EGFR a putative novel PCa related pathway [92]More recently an integrative analysis of genomic changesrevealed the role of the PI3K RASRAF and AR pathwaysin metastatic prostate cancers [6]The above insights providea blueprint for the design of novel pathway inhibitors intargeted therapies for prostate cancer

Thus far the pathway-based approach holds great promisefor cancer prediction However the known pathways corre-spond merely to a small fraction of somatic alterations Thealterations that have not been assigned to a definitive pathwayundermine the basis for a strictly pathway-centric markerdiscovery In addition the cross-talk between differentsignaling pathways further complicated the pathway-basedanalysis Thus biomarker discovery has to shift toward anintegrative network-based approach that accounts moreextensive genomic alteration

6 PCa-Specific Databases

High throughput research in PCa has led to vast amountsof comprehensive datasets There has been a growing desireto integrate specific data types into a centralized databaseand make them publicly available Considerable efforts wereundertaken and to date various national multicenter and


institutional databases in the context of prostate cancer re-search are available Prostate gene database (PGDB httpwwwurogeneorgpgdb) is a curated database on genes orgenomic loci related to human prostate and prostatic diseases[93] Another database prostate expression database (PEDBhttpwwwpedborg) is a curated database that containstools for analyzing prostate gene expression in both cancerousand normal conditions [94] Dragon database of genesassociated with prostate cancer (DDPC httpcbrckaustedusaddpc) [95] is an integrated knowledge database thatprovides a multitude of information related to PCa and PCa-related genes ChromSorter [96] collects PCa chromosomalregions associated with human prostate cancer PCaMDB isa genotype-phenotype database that collects prostate cancerrelated gene and protein variants from published literaturesThese specific databases tend to include large numbers ofpatients from different geographic regions Their general-izability and statistical power offer researchers a uniqueopportunity to conduct prostate cancer research in variousareas

7 Translational Bioinformatics in PCaA Future Direction

Recent advances in NGS technologies have resulted in hugesequence datasets This poses a tremendous challenge for theemerging field of translational bioinformatics Translationalbioinformatics by definition is the development of storageanalytic and interpretive methods to optimize the transla-tion from bench (laboratory-based genomic discoveries) tobedside (evidence-based clinical applications) The aim oftranslational bioinformatics is to combine the innovationsand resources across the entire spectrum of translationalmedicine towards the betterment of humanhealth To achievethis goal the fundamental aspects of bioinformatics (egbioinformatics imaging informatics clinical informatics andpublic health informatics) need to be integrated (Figure 2)

As the first aspect bioinformatics is concerned withapplying computational approaches to comprehend the intri-cate biological details elucidating molecular and cellular pro-cesses of cancer Imaging informatics is focused on what hap-pens at the level of tissues and organs and informatics tech-niques are used for image interpretation Clinical bioinfor-matics focuses on data from individual patients It is orientedto provide the technical infrastructure to understand clinicalrisk factors and differential response to treatment at theindividual levels As for public health informatics the strat-ified population of patients is at the center of interest Infor-matics solutions are required to study shared genetic envi-ronmental and life style risk factors on a population level Inorder to achieve the technical and semantic interoperabilityof multidimensional data fundamental issues in informationexchange and repository are to be addressed

8 Future Perspectives

The prospects for NGS-based biomarkers are excellentHowever compared with array-based studies real in-depth

NGS is still costly and also the analysis pipeline is lessestablished thus challenging the use of NGS A survey inthe GEO indicated that NGS is currently finding modestapplication in the identification of PCa markers Table 1listed the number of published GEO series on human PCausingNGS versusmicroarraysTherefore further work is stillrequired beforeNGS can be routinely used in the clinic Issuesregarding data management data integration and biologicalvariation will have to be tackled

81 NGS in the Cloud The dramatic increase in sequenceroutput has outpaced the improvements in computationalinfrastructure necessary to process the huge volumes of dataFortunately an alternative computing architecture cloudcomputing has recently emerged which provides fast andcost-effective solutions to the analysis of large-scale sequencedata In cloud computing high parallel tasks are run ona computer cluster through a virtual operating system (orldquocloudrdquo) Underlying the clouds are compact and virtual-ized virtual machines (VMs) hosting computation-intensiveapplications from distributed users Cloud computing allowsusers to ldquorentrdquo processing power and storage virtually on theirdemand and pay for what they use There has been con-siderable enthusiasm in the bioinformatics community foruse of cloud computing services Figure 3 shows a schematicdrawing of the cloud-based NGS analysis

Recently exploratory efforts have been made in cloud-based DNA sequence storage OrsquoConnor et al [97] createdSeqWare Query Engine using cloud computing technologiesto support databasing and query of information from thou-sands of genomes BaseSpace is a scalable cloud-computingplatform for all of Illuminarsquos sequencing systems After asequencing run is completed data from sequencing instru-ments is automatically uploaded toBaseSpace for analysis andstorage DNAnexus a company specialized in Cloud-basedDNA data analysis has leveraged the storage capacity ofGoogle Cloud to provide high-performance storage for NGSdata

Some initiatives have utilized preconfigured softwareon such cloud systems to process and analyze NGS dataTable 3 summarizes some tools that are currently available forsequence alignment short read mapping single nucleotidepolymorphism (SNP) identification and RNA expressionanalysis amongst others

Although cloud computing seems quite attractive thereare also issues that are yet to be resolvedThemost significantconcerns pertain to information security and bandwidth lim-itation Transferring massive amounts of data (on the orderof petabytes) to the cloud may be time consuming andprohibitively expensive For most sequencing centers thatrequire substantial data movement on a regular basis cloudcomputing currently does notmake economic senseWhile itis clear that cloud computing has great potential for researchpurposes for small labs and clinical applications using bench-top genome sequencers of limited throughput like MiSeqPGM and Proton cloud computing does have some practicalutility


Bioinformatics

Molecules

Translational biomedical

knowledgebase

Imaging informatics

Tissues and organs

Clinical informatics

Individuals

Public health informatics

Populationsociety

Figure 2 Translational bioinformatics bridges knowledge frommolecules to populations Four subdisciplines of translational bioinformaticsand their respective focus areas are depicted in boxesThe success of translational bioinformatics will enable a complete information logisticschain from single molecules to the entire human population and thus link innovations from bench to bedside

Cloud-based web services

Data storage

Master virtual machine

Worker nodes

Researcher

Sequencing lab Job queue

Data partition

Data

Figure 3 Schematic of the cloud-based NGS analysis Local computers allocate the cloud-based web services over the internet Web servicescomprised a cluster of virtual machines (one master node and a chosen number of worker nodes) Input data are transferred to the cloudstorage and the program code driving the computation is uploaded to master nodes by which worker nodes are provisioned Each workernode downloads reads from the storage and run computation independentlyThe final result is stored and meanwhile transferred to the localclient computer and the job completes


Table 3 The cloud computing software for NGS data analysis

Software Website Description ReferencesCrossbow httpbowtie-biosourceforgenetcrossbow Read mapping and SNP calling [56]CloudBurst httpcloudburst-biosourceforgenet Reference-based read mapping [57]Contrail httpcontrail-biosourceforgenet De novo read assembly [58]Cloud-MAQ httpsourceforgenetprojectscloud-maq Read mapping and assembly [59]Bioscope httpwwwlifescopecloudcom Reference-based read mapping [60]

GeneSifter httpwwwgeospizacomProductsAnalysisEditionshtml Customer oriented NGS data analysisservices [61]

CloudAligner httpsourceforgenetprojectscloudaligner Read mapping [62]

Roundup httprodeomedharvardedutoolsroundup Optimized computation for comparativegenomics [63]

PeakRanger httpwwwmodencodeorgsoftwareranger Peak caller for ChIP-Seq data [64]

Myrna httpbowtie-biosfnetmyrna Differential expression analysis forRNA-Seq data [65]

ArrayExpressHTS httpwwwebiacukToolsrwiki RNA-Seq data processing and qualityassessment [66]

SeqMapreduce Not available Read mapping [67]BaseSpace httpsbasespaceilluminacomhomeindex

82 Biomarker Discovery Using Systems Biology ApproachNGS makes it possible to generate multiple types of genomicalterations including mutations gene fusions copy num-ber alterations and epigenetic changes simultaneously ina single test Integration of these genomic transcriptomicinteractomic and epigenomic pieces of information is essen-tial to infer the underlying mechanisms in prostate cancerdevelopment The challenge ahead will be developing acomprehensive approach that could be analysed across thesecomplementary data looking for an ideal combination ofbiomarker signatures

Consequently the future of biomarker discovery will relyon a systems biology approach One of the most fascinatingfields in this regard is the network-based approaches tobiomarker discovery which integrate a large and heteroge-neous dataset into interactive networks In such networksmolecular components and interactions between them arerepresented as nodes and edges respectively The knowledgeextracted from different types of networks can assist discov-ery of novel biomarkers for example functional pathwaysprocessesor subnetworks for improved diagnostic prognos-tic and drug response prediction Network-based discoveryframework has already been reported in several types ofcancers [98ndash102] including PCa [103 104] In a pioneeringstudy Jin et al [103] built up a prostate-cancer-related net-work (PCRN) by searching the interactions among identifiedmolecules related to prostate cancerThe network biomarkersderived from the network display high-performances in PCapatient classification

As the field high-throughput technologies continues todevelop we will expect enhancing cooperation among differ-ent disciplines in translational bioinformatics such as bioin-formatics imaging informatics clinical informatics and pub-lic health informatics Notably the heterogeneous data typescoming from various informatics platforms are pushing for

developing standards for data exchange across specializeddomains

83 Personalized Biomarkers The recent breakthroughs inNGS also promise to facilitate the area of ldquopersonalisedrdquobiomarkers Prostate cancer is highly heterogeneous amongindividuals Current evidence indicates that the inter-individual heterogeneity arises from genetical environmentaland lifestyle factors By deciphering the genetic make-upof prostate tumors NGS may facilitate patient stratificationfor targeted therapies and therefore assist tailoring the besttreatment to the right patient It is envisioned that personal-ized therapy will become part of clinical practice for prostatecancer in the near future

9 Conclusions

Technological advances in NGS have increased our knowl-edge in molecular basis of PCa However the translation ofmultiple molecular markers into the clinical realm is in itsearly stages The full application of translational bioinfor-matics in PCa diagnosis and prognosis requires collaborativeefforts between multiple disciplines We can envision thatthe cloud-supported translational bioinformatics endeavourswill promote faster breakthroughs in the diagnosis andprognosis of prostate cancer

Conflict of Interests

The authors declare that there is no conflict of interests

Authorsrsquo Contribution

Jiajia Chen Daqing Zhang and Wenying Yan contributedequally to this work


Acknowledgments

The authors gratefully acknowledge financial support fromthe National Natural Science Foundation of China Grants(91230117 31170795) the Specialized Research Fund forthe Doctoral Program of Higher Education of China(20113201110015) International SampTCooperation Program ofSuzhou (SH201120) the National High Technology Researchand Development Program of China (863 program Grantno 2012AA02A601) and Natural Science Foundation forColleges and Universities in Jiangsu Province (13KJB180021)

References

[1] S Carlsson A J Vickers M Roobol et al ldquoProstate cancerscreening facts statistics and interpretation in response tothe us preventive services task force reviewrdquo Journal of ClinicalOncology vol 30 no 21 pp 2581ndash2584 2012

[2] LD F Venderbos andM J Roobol ldquoPSA-based prostate cancerscreening the role of active surveillance and informed andshared decisionmakingrdquoAsian Journal of Andrology vol 13 no2 pp 219ndash224 2011

[3] O Sartor ldquoRandomized studies of PSA screening an opinionrdquoAsian Journal of Andrology vol 13 no 3 pp 364ndash365 2011

[4] I N Sarkar ldquoBiomedical informatics and translationalmedicinerdquo Journal of Translational Medicine vol 8 article 222010

[5] C A Kulikowski and CW Kulikowski ldquoBiomedical and healthinformatics in translational medicinerdquo Methods of Informationin Medicine vol 48 no 1 pp 4ndash10 2009

[6] B S Taylor N Schultz H Hieronymus et al ldquoIntegrativegenomic profiling of human prostate cancerrdquo Cancer Cell vol18 no 1 pp 11ndash22 2010

[7] W Liu S Laitinen S Khan et al ldquoCopy number analysis indi-cates monoclonal origin of lethal metastatic prostate cancerrdquoNature Medicine vol 15 no 5 pp 559ndash565 2009

[8] X Mao Y Yu L K Boyd et al ldquoDistinct genomic alterationsin prostate cancers in Chinese and Western populations sug-gest alternative pathways of prostate carcinogenesisrdquo CancerResearch vol 70 no 13 pp 5207ndash5212 2010

[9] T W Friedlander R Roy S A Tomlins et al ldquoCommonstructural and epigenetic changes in the genome of castration-resistant prostate cancerrdquo Cancer Research vol 72 no 3 pp616ndash625 2012

[10] J SunW Liu T S Adams et al ldquoDNA copy number alterationsin prostate cancers a combined analysis of published CGHstudiesrdquo Prostate vol 67 no 7 pp 692ndash700 2007

[11] C M Robbins W A Tembe A Baker et al ldquoCopy numberand targeted mutational analysis reveals novel somatic eventsin metastatic prostate tumorsrdquo Genome Research vol 21 no 1pp 47ndash55 2011

[12] M J Kim R D Cardiff N Desai et al ldquoCooperativity of Nkx31and Pten loss of function in a mouse model of prostate carcino-genesisrdquo Proceedings of the National Academy of Sciences of theUnited States of America vol 99 no 5 pp 2884ndash2889 2002

[13] J Luo S Zha W R Gage et al ldquo120572-methylacyl-CoA racemasea new molecular marker for prostate cancerrdquo Cancer Researchvol 62 no 8 pp 2220ndash2226 2002

[14] M TinzlMMarberger S Horvath andC Chypre ldquoDD3PCA3RNAanalysis in urinemdashanewperspective for detecting prostatecancerrdquo European Urology vol 46 no 2 pp 182ndash187 2004

[15] E S Leman G W Cannon B J Trock et al ldquoEPCA-2 a highlyspecific serum marker for prostate cancerrdquo Urology vol 69 no4 pp 714ndash720 2007

[16] J Luo D J Duggan Y Chen et al ldquoHuman prostate cancerand benign prostatic hyperplasia molecular dissection by geneexpression profilingrdquo Cancer Research vol 61 no 12 pp 4683ndash4688 2001

[17] M F Darson A Pacelli P Roche et al ldquoHuman glandularkallikrein 2 (hK2) expression in prostatic intraepithelial neo-plasia and adenocarcinoma a novel prostate cancer markerrdquoUrology vol 49 no 6 pp 857ndash862 1997

[18] S Varambally S M Dhanasekaran M Zhou et al ldquoThepolycomb group protein EZH2 is involved in progression ofprostate cancerrdquo Nature vol 419 no 6907 pp 624ndash629 2002

[19] S F Shariat B Andrews M W Kattan J Kim T M Wheelerand KM Slawin ldquoPlasma levels of interleukin-6 and its solublereceptor are associated with prostate cancer progression andmetastasisrdquo Urology vol 58 no 6 pp 1008ndash1015 2001

[20] A Berruti A Mosca M Tucci et al ldquoIndependent prognosticrole of circulating chromogranin A in prostate cancer patientswith hormone-refractory diseaserdquo Endocrine-Related Cancervol 12 no 1 pp 109ndash117 2005

[21] S F Shariat C G Roehrborn J D McConnell et al ldquoAsso-ciation of the circulating levels of the urokinase system ofplasminogen activation with the presence of prostate cancerand invasion progression and metastasisrdquo Journal of ClinicalOncology vol 25 no 4 pp 349ndash355 2007

[22] S F Shariat M W Kattan E Traxel et al ldquoAssociation of pre-and postoperative plasma levels of transforming growth factor1205731 and interleukin 6 and its soluble receptor with prostatecancer progressionrdquo Clinical Cancer Research vol 10 no 6 pp1992ndash1999 2004

[23] S F Shariat J H Kim B Andrews et al ldquoPreoperative plasmalevels of transforming growth factor beta(1) strongly predictclinical outcome in patients with bladder carcinomardquo Cancervol 92 no 12 pp 2985ndash2992 2001

[24] J Lapointe C Li J P Higgins et al ldquoGene expression profilingidentifies clinically relevant subtypes of prostate cancerrdquo Pro-ceedings of the National Academy of Sciences of the United Statesof America vol 101 no 3 pp 811ndash816 2004

[25] G Kristiansen C Pilarsky C Wissmann et al ldquoExpressionprofiling of microdissected matched prostate cancer samplesreveals CD166MEMD and CD24 as new prognostic markersfor patient survivalrdquo Journal of Pathology vol 205 no 3 pp359ndash376 2005

[26] J Lapointe S Malhotra J P Higgins et al ldquohCAP-D3 expres-sion marks a prostate cancer subtype with favorable clinicalbehavior and androgen signaling signaturerdquo American Journalof Surgical Pathology vol 32 no 2 pp 205ndash209 2008

[27] K D Soslashrensen P J Wild A Mortezavi et al ldquoGenetic andepigenetic SLC18A2 silencing in prostate cancer is an indepen-dent adverse predictor of biochemical recurrence after radicalprostatectomyrdquoClinical Cancer Research vol 15 no 4 pp 1400ndash1410 2009

[28] J F Knight C J Shepherd S Rizzo et al ldquoTEAD1 and c-Cblare novel prostate basal cell markers that correlate with poorclinical outcome in prostate cancerrdquo British Journal of Cancervol 99 no 11 pp 1849ndash1858 2008

[29] W D Zhong G Q Qin Q S Dai et al ldquoSOXs in humanprostate cancer implication as progression and prognosisfactorsrdquo BMC Cancer vol 12 no 1 p 248 2012


[30] C Ruiz M Oeggerli M Germann et al ldquoHigh NRBP1 expres-sion in prostate cancer is linkedwith poor clinical outcomes andincreased cancer cell growthrdquo Prostate 2012

[31] N Pertega-Gomes J R Vizcaıno V Miranda-Goncalves et alldquoMonocarboxylate transporter 4 (MCT4) and CD147 overex-pression is associated with poor prognosis in prostate cancerrdquoBMC Cancer vol 11 p 312 2011

[32] A S Syed Khaja L Helczynski A Edsjo et al ldquoElevated levelofWnt5a protein in localized prostate cancer tissue is associatedwith better outcomerdquoPloSONE vol 6 no 10 Article ID e265392011

[33] M Q Yang B D Athey H R Arabnia et al ldquoHigh-throughputnext-generation sequencing technologies foster new cutting-edge computing techniques in bioinformaticsrdquo BMCGenomicsvol 10 no 1 article I1 2009

[34] C C Collins S V Volik A V Lapuk et al ldquoNext generationsequencing of prostate cancer from a patient identifies a defi-ciency of methylthioadenosine phosphorylase an exploitabletumor targetrdquo Molecular Cancer Therapeutics vol 11 no 3 pp775ndash783 2012

[35] C E Barbieri S C BacaM S Lawrence et al ldquoExome sequenc-ing identifies recurrent SPOP FOXA1 and MED12 mutationsin prostate cancerrdquo Nature Genetics vol 44 no 6 pp 685ndash6892012

[36] C S Grasso Y M Wu D R Robinson et al ldquoThe mutationallandscape of lethal castration-resistant prostate cancerrdquoNaturevol 487 no 7406 pp 239ndash243 2012

[37] A Kumar T AWhite A P MacKenzie et al ldquoExome sequenc-ing identifies a spectrum of mutation frequencies in advancedand lethal prostate cancersrdquo Proceedings of the NationalAcademy of Sciences of the United States of America vol 108 no41 pp 17087ndash17092 2011

[38] S A Tomlins D R Rhodes S Perner et al ldquoRecurrent fusionof TMPRSS2 and ETS transcription factor genes in prostatecancerrdquo Science vol 310 no 5748 pp 644ndash648 2005

[39] S A Tomlins R Mehra D R Rhodes et al ldquoTMPRSS2ETV4gene fusions define a third molecular subtype of prostatecancerrdquo Cancer Research vol 66 no 7 pp 3396ndash3400 2006

[40] B EHelgeson SA TomlinsN Shah et al ldquoCharacterization ofTMPRSS2ETV5 and SLC45A3ETV5 gene fusions in prostatecancerrdquo Cancer Research vol 68 no 1 pp 73ndash80 2008

[41] Z Shaikhibrahim M Braun P Nikolov et al ldquoRearrangementof the ETS genes ETV-1 ETV-4 ETV-5 and ELK-4 is a clonalevent during prostate cancer progressionrdquo Human Pathologyvol 43 no 11 pp 1910ndash1916 2012

[42] S A Tomlins B Laxman S M Dhanasekaran et al ldquoDistinctclasses of chromosomal rearrangements create oncogenic ETSgene fusions in prostate cancerrdquo Nature vol 448 no 7153 pp595ndash599 2007

[43] K G Hermans A A Bressers H A VanDer Korput N F DitsG Jenster and J Trapman ldquoTwo unique novel prostate-specificand androgen-regulated fusion partners of ETV4 in prostatecancerrdquo Cancer Research vol 68 no 9 pp 3094ndash3098 2008

[44] N Palanisamy B Ateeq S Kalyana-Sundaram et al ldquoRear-rangements of the RAF kinase pathway in prostate cancergastric cancer and melanomardquo Nature Medicine vol 16 no 7pp 793ndash798 2010

[45] C Wu A W Wyatt A V Lapuk et al ldquoIntegrated genome andtranscriptome sequencing identifies a novel form of hybrid andaggressive prostate cancerrdquo Journal of Pathology vol 227 no 1pp 53ndash61 2012

[46] H Beltran R Yelensky G M Frampton et al ldquoTargeted next-generation sequencing of advanced prostate cancer identifiespotential therapeutic targets and disease heterogeneityrdquo Euro-pean Urology vol 63 no 5 pp 920ndash926 2013

[47] K Kannan LWang J Wang MM IttmannW Li and L YenldquoRecurrent chimeric RNAs enriched in human prostate canceridentified by deep sequencingrdquo Proceedings of the NationalAcademy of Sciences of the United States of America vol 108 no22 pp 9172ndash9177 2011

[48] J R Prensner M K Iyer O A Balbin et al ldquoTranscriptomesequencing across a prostate cancer cohort identifies PCAT-1 an unannotated lincRNA implicated in disease progressionrdquoNature Biotechnology vol 29 no 8 pp 742ndash749 2011

[49] A Watahiki Y Wang J Morris et al ldquoMicroRNAs associatedwithmetastatic prostate cancerrdquo PLoS ONE vol 6 no 9 ArticleID e24950 2011

[50] K R Chng C W Chang S K Tan et al ldquoA transcriptionalrepressor co-regulatory network governing androgen responsein prostate cancersrdquo EMBO Journal vol 31 pp 2810ndash2823 2012

[51] Z Zhang CW ChangW L GohW-K Sung and E CheungldquoCENTDIST discovery of co-associated factors by motif distri-butionrdquo Nucleic Acids Research vol 39 no 2 pp W391ndashW3992011

[52] HHHe C AMeyerMW Chen V C JordanM Brown andX S Liu ldquoDifferential DNase I hypersensitivity reveals factor-dependent chromatin dynamicsrdquo Genome Research vol 22 no6 pp 1015ndash1025 2012

[53] G H Little H Noushmehr S K Baniwal B P Berman GA Coetzee and B Frenkel ldquoGenome-wide Runx2 occupancyin prostate cancer cells suggests a role in regulating secretionrdquoNucleic Acids Research vol 40 no 8 pp 3538ndash3547 2012

[54] P Y Tan C W Chang K R Chng K D Senali AbayratnaWansa W-K Sung and E Cheung ldquoIntegration of regulatorynetworks by NKX3-1 promotes androgen-dependent prostatecancer survivalrdquo Molecular and Cellular Biology vol 32 no 2pp 399ndash414 2012

[55] J H Kim S M Dhanasekaran J R Prensner et al ldquoDeepsequencing reveals distinct patterns of DNA methylation inprostate cancerrdquo Genome Research vol 21 no 7 pp 1028ndash10412011

[56] B Langmead M C Schatz J Lin M Pop and S L SalzbergldquoSearching for SNPs with cloud computingrdquo Genome Biologyvol 10 no 11 article R134 2009

[57] M C Schatz ldquoCloudBurst highly sensitive read mapping withMapReducerdquo Bioinformatics vol 25 no 11 pp 1363ndash1369 2009

[58] M C Schatz D D Sommer D R Kelley and M Pop ldquoDeNovo assembly of large genomes using cloud computingrdquo inProceedings of the Cold Spring Harbor Biology of GenomesConference New York NY USA May 2010

[59] A K Talukder S Gandham H A Prahalad and N PBhattacharyya ldquoCloud-MAQ the cloud-enabled scalablewholegenome reference assembly applicationrdquo in Proceedings of the7th IEEE and IFIP International Conference on Wireless andOptical Communications Networks (WOCN rsquo10) Colombo SriLanka September 2010

[60] M H Rahimi Bioscope Actuated Sensor Network for BiologicalScience University of Southern California Los Angeles CalifUSA 2005

[61] C Sansom ldquoUp in a cloudrdquo Nature Biotechnology vol 28 no1 pp 13ndash15 2010


[62] T NguyenW Shi andD Ruden ldquoCloudAligner a fast and full-featured MapReduce based tool for sequence mappingrdquo BMCResearch Notes vol 4 article 171 2011

[63] P Kudtarkar T F DeLuca V A Fusaro P J Tonellato andD P Wall ldquoCost-effective cloud computing a case studyusing the comparative genomics tool rounduprdquo EvolutionaryBioinformatics vol 2010 no 6 pp 197ndash203 2010

[64] X Feng R Grossman and L Stein ldquoPeakRanger a cloud-enabled peak caller for ChIP-seq datardquoBMCBioinformatics vol12 article 139 2011

[65] B Langmead K D Hansen and J T Leek ldquoCloud-scaleRNA-sequencing differential expression analysis with MyrnardquoGenome Biology vol 11 no 8 Article ID R83 2010

[66] A Goncalves A Tikhonov A Brazma and M Kapushesky ldquoApipeline for RNA-seq data processing and quality assessmentrdquoBioinformatics vol 27 no 6 pp 867ndash869 2011

[67] Y Li and S Zhong ldquoSeqMapReduce software and web servicefor accelerating sequence mappingrdquo Critical Assessment ofMassive Data Anaysis (CAMDA) vol 2009 2009

[68] M F Berger M S Lawrence F Demichelis et al ldquoThe genomiccomplexity of primary human prostate cancerrdquoNature vol 470no 7333 pp 214ndash220 2011

[69] I N Holcomb D I Grove M Kinnunen et al ldquoGenomicalterations indicate tumor origin and variedmetastatic potentialof disseminated cells from prostate cancer patientsrdquo CancerResearch vol 68 no 14 pp 5599ndash5608 2008

[70] Z Kan B S Jaiswal J Stinson et al ldquoDiverse somatic mutationpatterns andpathway alterations in human cancersrdquoNature vol466 no 7308 pp 869ndash873 2010

[71] L Agell S Hernandez S De Muga et al ldquoKLF6 and TP53mutations are a rare event in prostate cancer distinguishingbetween Taq polymerase artifacts and true mutationsrdquoModernPathology vol 21 no 12 pp 1470ndash1478 2008

[72] J-T Dong ldquoPrevalent mutations in prostate cancerrdquo Journal ofCellular Biochemistry vol 97 no 3 pp 433ndash447 2006

[73] X Sun H F Frierson C Chen et al ldquoFrequent somaticmutations of the transcription factor ATBF1 in human prostatecancerrdquo Nature Genetics vol 37 no 6 pp 407ndash412 2005

[74] L Zheng F Wang C Qian et al ldquoUnique substitution ofCHEK2 and TP53 mutations implicated in primary prostatetumors and cancer cell linesrdquo Human Mutation vol 27 no 10pp 1062ndash1063 2006

[75] C Kumar-Sinha S A Tomlins and A M Chinnaiyan ldquoRecur-rent gene fusions in prostate cancerrdquo Nature Reviews Cancervol 8 no 7 pp 497ndash511 2008

[76] S Perner J-MMosquera F Demichelis et al ldquoTMPRSS2-ERGfusion prostate cancer an early molecular event associated withinvasionrdquo American Journal of Surgical Pathology vol 31 no 6pp 882ndash888 2007

[77] S A Tomlins S M J Aubin J Siddiqui et al ldquoUrineTMPRSS2ERG fusion transcript stratifies prostate cancerrisk in men with elevated serum PSArdquo Science TranslationalMedicine vol 3 no 94 Article ID 94ra72 2011

[78] D B Seligson SHorvath T Shi et al ldquoGlobal histonemodifica-tion patterns predict risk of prostate cancer recurrencerdquoNaturevol 435 no 7046 pp 1262ndash1266 2005

[79] A Urbanucci B Sahu J Seppala et al ldquoOverexpression ofandrogen receptor enhances the binding of the receptor to thechromatin in prostate cancerrdquoOncogene vol 31 no 17 pp 2153ndash2163 2012

[80] D S Rickman T D Soong BMoss et al ldquoOncogene-mediatedalterations in chromatin conformationrdquo Proceedings of theNational Academy of Sciences of the United States of Americavol 109 no 23 pp 9083ndash9088 2012

[81] X Lin M Tascilar W-H Lee et al ldquoGSTP1 CpG island hyper-methylation is responsible for the absence of GSTP1 expressionin human prostate cancer cellsrdquo American Journal of Pathologyvol 159 no 5 pp 1815ndash1826 2001

[82] E Rosenbaum M O Hoque Y Cohen et al ldquoPromoter hyper-methylation as an independent prognostic factor for relapse inpatients with prostate cancer following radical prostatectomyrdquoClinical Cancer Research vol 11 no 23 pp 8321ndash8325 2005

[83] O Taiwo G A Wilson T Morris et al ldquoMethylome analysisusingMeDIP-seq with low DNA concentrationsrdquoNature Proto-cols vol 7 no 4 pp 617ndash636 2012

[84] R Lister M Pelizzola R H Dowen et al ldquoHuman DNAmethylomes at base resolution show widespread epigenomicdifferencesrdquo Nature vol 462 no 7271 pp 315ndash322 2009

[85] A Meissner A Gnirke G W Bell B Ramsahoye E S Landerand R Jaenisch ldquoReduced representation bisulfite sequencingfor comparative high-resolution DNA methylation analysisrdquoNucleic Acids Research vol 33 no 18 pp 5868ndash5877 2005

[86] C Gebhard L Schwarzfischer T-H Pham et al ldquoGenome-wide profiling of CpG methylation identifies novel targetsof aberrant hypermethylation in myeloid leukemiardquo CancerResearch vol 66 no 12 pp 6118ndash6128 2006

[87] A L Brunner D S Johnson W K Si et al ldquoDistinct DNAmethylation patterns characterize differentiated human embry-onic stem cells and developing human fetal liverrdquo GenomeResearch vol 19 no 6 pp 1044ndash1056 2009

[88] M W Kattan S F Shariat B Andrews et al ldquoThe additionof interleukin-6 soluble receptor and transforming growthfactor beta1 improves a preoperative nomogram for predictingbiochemical progression in patients with clinically localizedprostate cancerrdquo Journal of Clinical Oncology vol 21 no 19 pp3573ndash3579 2003

[89] J Baden G Green J Painter et al ldquoMulticenter evaluation ofan investigational prostate cancer methylation assayrdquo Journal ofUrology vol 182 no 3 pp 1186ndash1193 2009

[90] D R Rhodes T R Barrette M A Rubin D Ghosh and A MChinnaiyan ldquoMeta-analysis of microarrays interstudy valida-tion of gene expression profiles reveals pathway dysregulationin prostate cancerrdquo Cancer Research vol 62 no 15 pp 4427ndash4433 2002

[91] G V Glinsky A B Glinskii A J Stephenson R M Hoffmanand W L Gerald ldquoGene expression profiling predicts clinicaloutcome of prostate cancerrdquo Journal of Clinical Investigationvol 113 no 6 pp 913ndash923 2004

[92] Y Wang J Chen Q Li et al ldquoIdentifying novel prostate cancerassociated pathways based on integrative microarray data anal-ysisrdquo Computational Biology and Chemistry vol 35 no 3 pp151ndash158 2011

[93] L-C Li H ZhaoH Shiina C J Kane andRDahiya ldquoPGDB acurated and integrated database of genes related to the prostaterdquoNucleic Acids Research vol 31 no 1 pp 291ndash293 2003

[94] V Hawkins D Doll R Bumgarner et al ldquoPEDB the prostateexpression databaserdquo Nucleic Acids Research vol 27 no 1 pp204ndash208 1999

[95] M Maqungo M Kaur S K Kwofie et al ldquoDDPC dragondatabase of genes associated with prostate cancerrdquoNucleic AcidsResearch vol 39 no 1 pp D980ndashD985 2011


[96] A Etim G Zhou XWen et al ldquoChromSorter PC a database ofchromosomal regions associated with human prostate cancerrdquoBMC Genomics vol 5 article 27 2004

[97] BDOrsquoConnor BMerriman and S F Nelson ldquoSeqWareQueryEngine storing and searching sequence data in the cloudrdquo BMCBioinformatics vol 11 no 12 article S2 2010

[98] H-Y Chuang E Lee Y-T Liu D Lee and T Ideker ldquoNetwork-based classification of breast cancer metastasisrdquo MolecularSystems Biology vol 3 article 140 2007

[99] IW Taylor R Linding DWarde-Farley et al ldquoDynamicmod-ularity in protein interaction networks predicts breast canceroutcomerdquo Nature Biotechnology vol 27 no 2 pp 199ndash2042009

[100] MXuM-C J Kao J Nunez-Iglesias J R NevinsMWest andX J Jasmine ldquoAn integrative approach to characterize disease-specific pathways and their coordination a case study in cancerrdquoBMC Genomics vol 9 no 1 article S12 2008

[101] Y-C Wang and B-S Chen ldquoA network-based biomarkerapproach for molecular investigation and diagnosis of lungcancerrdquo BMCMedical Genomics vol 4 article 2 2011

[102] Y Zhang S Wang D Li et al ldquoA systems biology-based clas-sifier for hepatocellular carcinoma diagnosisrdquo PLoS ONE vol6 no 7 Article ID e22426 2011

[103] G Jin X Zhou K Cui X-S Zhang L Chen and S T CWong ldquoCross-platform method for identifying candidate net-work biomarkers for prostate cancerrdquo IET Systems Biology vol3 no 6 pp 505ndash512 2009

[104] R Ummanni F Mundt H Pospisil et al ldquoIdentification ofclinically relevant protein targets in prostate cancer with 2D-DIGE coupledmass spectrometry and systems biology networkplatformrdquo PLoS ONE vol 6 no 2 Article ID e16833 2011

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Anatomy Research International

PeptidesInternational Journal of


Hindawi Publishing Corporation httpwwwhindawicom

International Journal of

Volume 2014

Zoology


Molecular Biology International

GenomicsInternational Journal of


The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


BioinformaticsAdvances in

Marine BiologyJournal of



Signal TransductionJournal of


BioMed Research International

Evolutionary BiologyInternational Journal of



Biochemistry Research International

ArchaeaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Genetics Research International


Advances in

Virolog y

Hindawi Publishing Corporationhttpwwwhindawicom

Nucleic AcidsJournal of

Volume 2014

Stem CellsInternational



Enzyme Research



Microbiology


Table 1 Number of PCa-associated GEO series generated by microarray and NGS

Methodology Gene expressionprofiling

Noncoding RNAprofiling

Genomebindingoccupancy

profiling

Genome methylationprofiling

Genome variationprofiling

Microarray 266 34 17 21 35NGS 11 1 18 2 2

array-based profiles in human PCaThe retrieved GEO seriesgenerally fall into 5 categories gene expression profilingnoncoding RNA profiling genome bindingoccupancy pro-filing genome methylation profiling and genome variationprofiling The number of GEO series for each category issummarized in Table 1

Together these array-based technologies have shed lighton the genetic alterations in the PCa genome Among theabnormalities affecting prostate tumors the copy numberalteration is the most common one [6]

Numerous early studies have used comparative genomehybridization (CGH) or single nucleotide polymorphism(SNP) arrays to assess copy number changes in tumor DNAAs a result multiple genomic regions that displayed frequentgain or loss in the PCa genome [6ndash11] have been revealedChromosome 3p14 8p22 10q23 13q13 and 13q14 are foundto display broad copy number deletion Key genes mappingwithin these deleted regions include NKX31 PTEN [12]BRCA2 C13ORF15 SIAH3 [11] RB1 HSD17B2 [9] FOXP1RYBP and SHQ1 [6] High-level copy number gains aredetected at 5p13 14q21 7q22 Xq12 and 8q13 [9] Keyamplified genes mapping within these regions include SKP2FOXA1 AR [11] and HSD17B3 [9]

Usingmicroarray substantial efforts have also beenmadeto characterize prostate cancer gene expression profilesDifferentially expressed genes identified in these studies pointto a plethora of candidate biomarkers with diagnostic orprognostic value

Adiagnosticmarker is able to differentiate prostate cancerwith other prostatic abnormalities There are many emergingmarkers that show promise for PCa diagnosis such as alpha-methylacyl-CoA racemase (AMACR) [13] prostate cancergene 3 (PCA3) [14] early prostate cancer antigen (EPCA)-2[15] Hepsin [16] kallikrein-related peptidase 2 (KLK2) [17]and polycomb group protein enhancer of zeste homolog 2(EZH2) [18] The most prominent of these is PCA3 whichwas found to exhibit higher sensitivity and specificity forPCa detection than PSA PCA3 thus provides a potentialcomplement to PSA for the early diagnosis of PCa

Prognostic biomarkers for prediction of prostate cancerpatient outcome have also been identified Increased serumlevels of IL-6 and its receptor (IL-6R) are associated withmetastatic and hormone refractory disease [19] Elevatedserum chromogranin A levels are indicative of poor progno-sis and decreased survival [20] Other differentially expressedmolecules with prognostic potential include the urokinaseplasminogen activation (uPA) [21] TGF-1205731 [22 23] MUC1[24] CD24 [25] hCAP-D3 [26] vesicular monoamine trans-porter 2 (SLC18A2) [27] TEA domain family member 1

(TEAD1) c-Cbl [28] SOX7 and SOX9 [29] nuclear receptorbinding protein 1 (NRBP1) [30] CD147 [31] and Wnt5a [32]Each of thesemarkers will require proper validation to ensuretheir clinical utility

While microarray technology represents a wonderfulopportunity for the detection of genomic alterations there aresignificant issues thatmust be considered For example array-based methods are impossible to detect variations at a lowfrequency (many well below 1) in the samples In additionmicroarrays can only provide information about the genesthat are already included on the array The emerging next-generation sequencing technology also called massively par-allel sequencing however helps to overcome the challengesby generating actual sequence reads [33]

3 NGS Based Diagnosis and Prognosis of PCa

Key feature of the next-generation sequencing technologyis the massive parallelization of the sequencing process Byvirtue of the massively parallel process NGS generates hun-dreds of millions of short DNA reads (100ndash250 nucleotides)simultaneously which are then assembled and aligned toreference genomes

A number of NGS systems are available commerciallyincluding Genome AnalyzerHiSeq 2000MiSeq from Illu-mina SOLiDPGMProton from Life Sciences GS-FLX(454)GS Junior from Roche as well as novel single moleculesequencers for example Heliscope from Helicos Biosciencesand SMRT offered by Pacific Biosciences All these technolo-gies provide digital information onDNA sequences andmakeit feasible to discover genetic mutations at unprecedentedresolution and lower cost

It is generally accepted that cancers are caused by theaccumulation of genomic alterations NGS methods can wellsqueeze all the alteration information that remains hiddenwithin the genome including point mutations small inser-tions and deletions (InDels) copy number alterations (CNV)chromosomal rearrangements and epigenetic alterationsFor these reasons NGS has become the method of choice forlarge-scale detection of somatic cancer genome alterationsand is changing the way how cancer genome is analyzed AnNGS-based research pipeline for PCa biomarkers is given inFigure 1

Currently next-generation sequencing is being applied tocancer genome study in various ways

(1) genome-based sequencing (DNA-Seq) yielding in-formation on sequence variation InDels chromoso-mal rearrangements and copy number variations






TFBS



Pathway databases

Statistical methods


Molecularalteration

Literatures

i d di1

1

2

2

3 4 5 6 7 8 9 10 11 12 13 14

G U A

mRNA

5998400

CG

GC

C

GGCMM

MM

Zfp198 Elys

Err2

Rif1

BAF155

Wdr18Btbd14a

Arid3aTif1B

Sall1

Pelo

Zfp609

Mybbp

NF45SP1

HDAC2

Cdk1

EWS

Zfp219

Rybp

RequiemArid3b

Wapl

P66B

Rex1

Oct4 Zfp281

Nanog Dax1

Nac1

Sall4

REST

Rnf2

Rai14

Prmt1 YY1

MAPK














DNA-Seq


RNA-Seq


Chip-Seq
















































Bioinformatics

Molecules


knowledgebase

Imaging informatics

Tissues and organs


Individuals


Populationsociety



Data storage


Worker nodes

Researcher


Data partition

Data

















9 Conclusions







Acknowledgments


References



















































































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology






TFBS



Pathway databases

Statistical methods


Molecularalteration

Literatures

i d di1

1

2

2

3 4 5 6 7 8 9 10 11 12 13 14

G U A

mRNA

5998400

CG

GC

C

GGCMM

MM

Zfp198 Elys

Err2

Rif1

BAF155

Wdr18Btbd14a

Arid3aTif1B

Sall1

Pelo

Zfp609

Mybbp

NF45SP1

HDAC2

Cdk1

EWS

Zfp219

Rybp

RequiemArid3b

Wapl

P66B

Rex1

Oct4 Zfp281

Nanog Dax1

Nac1

Sall4

REST

Rnf2

Rai14

Prmt1 YY1

MAPK














DNA-Seq


RNA-Seq


Chip-Seq
















































Bioinformatics

Molecules


knowledgebase

Imaging informatics

Tissues and organs


Individuals


Populationsociety



Data storage


Worker nodes

Researcher


Data partition

Data

















9 Conclusions







Acknowledgments


References



















































































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology




DNA-Seq


RNA-Seq


Chip-Seq
















































Bioinformatics

Molecules


knowledgebase

Imaging informatics

Tissues and organs


Individuals


Populationsociety



Data storage


Worker nodes

Researcher


Data partition

Data

















9 Conclusions







Acknowledgments


References



















































































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology










































Bioinformatics

Molecules


knowledgebase

Imaging informatics

Tissues and organs


Individuals


Populationsociety



Data storage


Worker nodes

Researcher


Data partition

Data

















9 Conclusions







Acknowledgments


References



















































































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology
















9 Conclusions







Acknowledgments


References



















































































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology


Acknowledgments


References



















































































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology






















































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology

review article translational bioinformatics for diagnostic

Documents