Microarrays and CancerSegal et al.CS 498 SSSaurabh SinhaGenomics and pathology• Genomics provides high-throughputmeasurements of molecular mechanisms– Microarrays, ChIP-on-chip, etc.• Genomics may provide the molecularunderpinnings of pathology, in a highlycomprehensive manner– Revolutionize the diagnosis and management ofdiseases, including cancerPrior applications to cancer• Gene expression measurements have beenapplied to cancer diagnosis• Measure each gene’s expression in severalnormal tissue samples, and severalpathological (diseased) samples• Find subset of genes differentially expressedin the two sample groups• If such “gene signatures” of particular cancertypes are found, they can become the basisof tests for malignancyWe want better …• Genes may be differentially expressed, butnot enough to cross certain thresholds usedin the analysis• Analyzing the data on a gene-by-gene basisis error prone -- microarray data has inherentnoise• Finding the genes involved in one type ofcancer is only the first step; it does not revealthe underlying processesPart 1: Cancer modulesA “module” level view• Many methods use “gene modules”(sets of genes) as basic blocks foranalysis• Instead of trying to find changes inindividual gene expression profiles, lookout for entire sets of genes withchanging expression profilesThe study of Mootha et al.• Showed that expression of “oxidativephosphorylation” genes (a particular setof genes) is reduced in diabetic muscle• Signal not very strong when looking atindividual genes, but highly significantwhen looking at the “gene module”Source: Nature Genetics 3 7, S38 - S45 (2005) Disease tissue(Diabetes mellitus type 2)Normal tissue(Normal tolerance to glucose)Grey: all genesRed: oxidative phosphorylation genesSegal et al.: Methodology• Compile a large collection of cancer-relatedmicroarrays– microarrays measuring gene expression in cancer tissues ornormal tissue• Compile a large collection of gene sets (modules)from earlier studies• Identify gene set (modules) induced or repressed in amicroarray• Identify modules induced in several arrays, orrepressed in several arrays• Check if these arrays are enriched in some clinicalannotationSegal et al: Cancer “module maps”Source: Nature Genetics 3 7, S38 - S45 (2005) Red(m,c): Microarrays in whichmodule m was overexpressed (induced) are enriched in condition cGreen: Microarrays in whichmodule m was underexpressed (repressed) are enriched in condition cRows and columns are not inan arbitrary order. They havebeen “clustered” to displaysimilar rows (or columns) togetherInsights from cancer module map• Some modules activated or repressedacross many tumor types. Suchmodules could be related to generaltumorogenic processes• Some modules specifically activated orrepressed in certain tumor types orstages of tumor progressionFrom modules to regulation• A module map shows the transcriptionalchanges underlying cancer• Transcriptional changes are a result oftranscription factors and their binding sites• A deeper understanding of cancer wouldcome from finding out which transcriptionfactors and binding sites led to thetranscriptional changesPart 2: Cis-regulatory elementsGenomics and gene regulation• Such knowledge comes from genomics data• ChIP-chip studies identify which transcriptionfactors bind which DNA sequences• Analysis of DNA sequence, using knownbinding site motifs, gives us putative bindingsites• Cross-species conservation also tells ussomething about possible locations of bindingsitesCis-regulatory analysis• Identify a set of genes whose promoterscontain the same binding sites– Such a set of genes is likely to be regulated by thesame TF– Often called a “regulatory module”• Earlier studies mined microarrays for “co-expressed” genes, then used motif findingalgorithms to discover their shared bindingsitesCis-regulatory analysis• Another approach (Segal et al.) tried to solvethe problem in an integrated manner• Find a set of genes such that– their expression profiles are similar (microarrays)– they share the same binding sites (sequence)• Joint learning of “regulatory module” from twovery different types of data: microarray andsequence– An important theme in current bioinformaticsCis-regulatory analysis• Connection between gene expressionand cis-regulatory elements (bindingsites) also explored in Beer & Tavazoie.• Found rules on combinations andlocations of binding sites that wouldcause the gene to be over- or under-expressed• The binding sites “RRPE” and “PAC” must occur within240 bp and 140 bp of gene start• Genes containing both motifs, following certain rules onlocation, are tightly co-regulated• Genes containing any one motif, or both in incorrectpositional configuration, have close to randomexpressionSource: Nature Genetics 3 7, S38 - S45 (2005)Eukaryotes• These studies have mostly focused on yeast(which is a eukaryote, but has a small,compact genome)• Not much work of this type in the longer,more complex genomes of metazoans (e.g.,humans, rodents, fruitflies)• The genome is not compact; may not sufficeto look at sequence right next to a gene.Intergenic regions are long, and cis-regulatorysignals may not be close to geneOne study in humans• HeLa cells are an “immortal” cell-linederived from cervical cancer cells in aperson who died in 1951.– Used extensively in studying cancer• Method of Segal et al. (joint learning ofregulatory modules from geneexpression and sequence data) appliedto these cellsOne study in humans• Gene expression data used:microarrays measuring genes duringcell cycle in HeLa cells• Sequence: 1000 bp promoters(upstream) of human genesResult of analysis: Two motifs found to be shared by this set of genes. The genes have similar expression profiles. One of theidentified motifs (NFAT) known to be involved in cell-cycleSource: Nature Genetics 3 7, S38 - S45 (2005)Summary• The common theme is to analyze setsof genes, and relate their commonexpression patterns to cancer types orto presence of cis-regulatory motifs• Search algorithms may be required toidentify some of these
View Full Document
Unlocking...