METHODS IN ENZYMOLOGY EDITORS-IN-CHIEF John N. Abelson Melvin I. Simon DIVISIONOFBIOLOGY CALIFORNIAINSTITUTEOFTECHNOLOGY PASADENA,CALIFORNIA FOUNDING EDITORS Sidney P. Colowick and Nathan O. Kaplan Table of Contents CONTRIBUTORSTOVOLUME411. . . . . . . . . . . . . . . . . ix VOLUMESINSERIES. . . . . . . . . . . . . . . . . . . . . xiii 1. RNAExtractionforArrays LAKSHMIV.MADABUSI, GARYJ.LATHAM,AND BERNARDF.ANDRUSS 1 2. AnalyzingMicro-RNAExpressionUsing TIMOTHYS.DAVISON, Microarrays CHARLESD.JOHNSON,AND BERNARDF.ANDRUSS 14 3. TroubleshootingMicroarrayHybridizations BRIANEADS, AMYCASH, KEVINBOGART, JAMESCOSTELLO,AND JUSTENANDREWS 34 4. UseofExternalControlsinMicroarray IVANAV.YANG 50 Experiments 5. StandardsinGeneExpressionMicroarray MARCSALIT 63 Experiments 6. ScanningMicroarrays:CurrentMethodsand JERILYNA.TIMLIN 79 FutureDirections 7. AnIntroductiontoBioArraySoftware CARLTROEIN,JOHAN Environment VALLON-CHRISTERSSON, ANDLAOH.SAAL 99 8. Bioconductor:AnOpenSourceFramework MARKREIMERSAND forBioinformaticsandComputational VINCENTJ.CAREY 119 Biology v vi TABLE OF CONTENTS 9. TM4MicroarraySoftwareSuite ALEXANDERI.SAEED, NIRMALK.BHAGABATI, JOHNC.BRAISTED, WEILIANG, VASILYSHAROV, ELEANORA.HOWE, JIANWEILI, MATHANGITHIAGARAJAN, JOSEPHA.WHITE,AND JOHNQUACKENBUSH 134 10. ClusteringMicroarrayData JEREMYGOLLUBAND GAVINSHERLOCK 194 11. AnalysisofVarianceofMicroarrayData JULIENF.AYROLESAND GREGGIBSON 214 12. MicroarrayQualityControl JAMESM.MINOR 233 13. AnalysisofaMultifactorMicroarrayStudy TOMDOWNEY 256 UsingPartekGenomicsSolution 14. StatisticsforChIP-chipandDNase PETERC.SCACHERI, HypersensitivityExperimentson GREGORYE.CRAWFORD, NimbleGenArrays ANDSEANDAVIS 270 15. ExtrapolatingTraditionalDNAMicroarray THOMASE.ROYCE, StatisticstoTilingandProtein JOELS.ROZOWSKY, MicroarrayTechnologies NICHOLASM.LUSCOMBE, OLOFEMANUELSSON, HAIYUANYU, XIAOWEIZHU, MICHAELSNYDER,AND MARKB.GERSTEIN 282 16. RandomDataSetGenerationtoSupport DANIELQ.NAIMAN 312 MicroarrayAnalysis 17. UsingOntologiestoAnnotate PATRICIAL.WHETZEL, MicroarrayExperiments HELENPARKINSON,AND CHRISTIANJ.STOECKERT,JR 325 18. InterpretingExperimentalResultsUsing TIMBEISSBARTH 340 GeneOntologies 19. GeneExpressionOmnibus: TANYABARRETTAND MicroarrayDataStorage,Submission, RONEDGAR 352 Retrieval,andAnalysis TABLEOFCONTENTS vii 20. DataStorageandAnalysisinArrayExpress ALVISBRAZMA, MISHAKAPUSHESKY, HELENPARKINSON, UGISSARKINS,AND MOHAMMADSHOJATALAB 370 21. ClusteringMethodsforAnalyzingLarge JE´ROˆMEHENNETINAND DataSets:GonadDevelopment, MICHELBELLIS 387 AStudyCase 22. VisualizingNetworks GEORGEW.BELLAND FRANLEWITTER 408 23. RandomForestsforMicroarrays ADELECUTLERAND JOHNR.STEVENS 422 AUTHORINDEX . . . . . . . . . . . . . . . . . . . . . . 433 SUBJECTINDEX . . . . . . . . . . . . . . . . . . . . . . 461 Contributors to Volume 411 Articlenumbersareinparenthesesandfollowingthenameofcontributors. Affiliationslistedarecurrent. JUSTEN ANDREWS (3), Department of AMYCASH(3),DepartmentofBiology,In- Biology, Indiana University, Blooming- dianaUniversity,Bloomington,Indiana ton,Indiana JAMES COSTELLO (3), Drosophila Geno- BERNARD F. ANDRUSS (1, 2), Asuragen, mics Resource Center, Indiana Univer- Inc.,Austin,Texas sity,Bloomington,Indiana JULIEN F. AYROLES (11), Department of GREGORYE. CRAWFORD(14), Department Genetics, North Carolina State Univer- of Pediatrics, Institute for Genome sity,Raleigh,NorthCarolina Sciences and Policy, Duke University, Durham,NorthCarolina TANYABARRETT(19),NationalCenter for Biotechnology Information, National ADELECUTLER(23),DepartmentofMathe- LibraryofMedicine,NationalInstitutes maticsandStatistics,UtahStateUniver- ofHealth,Bethesda,Maryland sity,Logan,Utah TIMBEISSBARTH(18),TheWalterandEliza SEANDAVIS(14),CancerGeneticsBranch, Hall Institute of Medical Research, Bio- NationalHumanGenomeResearchInsti- informaticsGroup,Victoria,Australia tute,NationalInstitutesofHealth,Bethes- da,Maryland GEORGEW.BELL(22),Bioinformaticsand Research Computing, Whitehead Insti- TIMOTHYS.DAVISON(2),AsuragenDiscov- tute for Biomedical Research, Cam- eryServices,Asuragen,Inc.,Austin,Texas bridge,Massachusetts TOM DOWNEY (13), Partek Incorporated, MICHELBELLIS(21),CRBM–CNRS,Mont- SaintLouis,Missouri pellier,France BRIAN EADS (3),DepartmentofBiology, NIRMALK.BHAGABATI(9), TheInstitutefor IndianaUniversity,Bloomington,Indiana GenomicResearch,Rockville,Maryland RON EDGAR (19), National Center for KEVIN BOGART(3),DrosophilaGenomics Biotechnology Information, National Resource Center, Indiana University, LibraryofMedicine,Bethesda,Maryland Bloomington,Indiana OLOF EMANUELSSON (15), Department of JOHN C. BRAISTED (9), The Institute for MolecularBiophysicsand Biochemistry, GenomicResearch,Rockville,Maryland YaleUniversity,NewHaven,Connecticut ALVIS BRAZMA (20), European Bioinfor- MARK B. GERSTEIN (15), Department of maticsInstitute,WellcomeTrustGenome MolecularBiophysicsand Biochemistry, Campus, Hinxton, Cambridge, United YaleUniversity,NewHaven,Connecticut Kingdom GREGGIBSON(11), DepartmentofGenet- VINCENTJ.CAREY(8),HarvardUniversity, ics, North Carolina State University, Boston,Massachusetts Raleigh,NorthCarolina ix x CONTRIBUTORS TO VOLUME 411 JEREMYGOLLUB*(10),DepartmentofBio- HELENPARKINSON(17,20),EuropeanBio- chemistry, Stanford University Medical informatics Institute, Wellcome Trust School,Stanford,California Genome Campus, Hinxton, Cambridge, UnitedKingdom JE´ROˆME HENNETIN (21), CRBM–CNRS, Montpellier,France JOHN QUACKENBUSH (9), Department of Biostatistics and Computational Biol- ELEANOR A. HOWE (9), Department of ogy, Dana-Farber Cancer Institute Bos- Biostatistics and Computational Biol- ton,Massachusetts ogy,Dana-FarberCancerInstitute,Bos- ton,Massachusetts MARKREIMERS(8),NationalCancerInsti- tute,Bethesda,Maryland CHARLES D. JOHNSON (2), Asuragen Dis- covery Services, Asuragen, Inc., Austin, THOMASE. ROYCE(15),PrograminCom- Texas putational Biology and Bioinformatics, YaleUniversity,NewHaven,Connecticut MISHA KAPUSHESKY (20), European Bio- informatics Institute, Wellcome Trust JOEL S. ROZOWSKY (15), Department of Genome Campus, Hinxton, Cambridge, MolecularBiophysicsandBiochemistry, UnitedKingdom YaleUniversity,NewHaven,Connecticut GAtiRnY,TJe.xLaAsTHAM(1),Asuragen,Inc.,Aus- LAOH.SAAL(7),InstituteforCancerGen- etics, College of Physicians and Sur- FRAN LEWITTER (22), Bioinformatics and geons,ColumbiaUniversity,NewYork, Research Computing, Whitehead Insti- NewYork tute for Biomedical Research, Cam- bridge,Massachusetts ALEXANDER I. SAEED (9), Department of Bioinformatics, The Institute for Geno- JIANWEI LI (9), The Institute for Genomic micResearch,Rockville,Maryland Research,Rockville,Maryland MARC SALIT (5), Chemical Science and WEILIANG(9),TheInstituteforGenomic Technology Laboratory, National Insti- Research,Rockville,Maryland tute of Standards and Technology, Gaithersburg,Maryland NICHOLAS M. LUSCOMBE (15), European Bioinformatics Institute,WellcomeTrust UGIS SARKANS (20), European Bioinfor- Genome Campus, Hinxton, Cambridge, matics Institute, Wellcome Trust Gen- UnitedKingdom ome Campus, Hinxton, Cambridge, UnitedKingdom LAKSHMIV. MADABUSI (1),AsuragenDis- covery Services, Asuragen, Inc., Austin, PETER C. SCACHERI (14), Department of Texas Genetics,CaseWesternReserveUniver- sity,Cleveland,Ohio JAMESM.MINOR(12),AgilentTechnologies, Inc.,SantaClara,California VASILY SHAROV (9), The Institute for GenomicResearch,Rockville,Maryland DANIEL Q. NAIMAN (16), Department of Applied Mathematics and Statistics, GAVIN SHERLOCK (10), Department of Johns Hopkins University, Baltimore, Genetics, Stanford University Medical Maryland School,Stanford,California *Currentaffiliation:InconixPharmaceuticals,Inc.,MountainView,California. CONTRIBUTORSTO VOLUME411 xi MOHAMMAD SHOJATALAB (20), European JOHAN VALLON-CHRISTERSSON(7),Depart- BioinformaticsInstitute,WellcomeTrust ment of Oncology, Lund University Genome Campus, Hinxton, Cambridge, Hospital,Lund,Sweden UnitedKingdom PATRICIA L. WHETZEL (17), Department MICHAEL SNYDER(15),DepartmentofMo- of Genetics, Center for Bioinformatics, lecular, Cellular, and Developmental University of Pennsylvania School of Biology, Yale University, New Haven, Medicine,Philadelphia,Pennsylvania Connecticut JOSEPHA.WHITE(9),DepartmentofBios- JOHNR.STEVENS(23),DepartmentofMath- tatistics and Computational Biology, ematicsandStatistics,UtahStateUniver- Dana-Farber Cancer Institute, Boston, sity,Logan,Utah Massachusetts CHRISTIANJ. STOECKERT,JR.(17), Depart- mentofGenetics,CenterforBioinforma- IVANA V. YANG (4), Laboratory of Res- piratory Biology, National Institute of tics,UniversityofPennsylvaniaSchoolof Environmental Health Sciences, Re- Medicine,Philadelphia,Pennsylvania searchTrianglePark,NorthCarolina MATHANGI THIAGARAJAN (9), Department of Bioinformatics, The Institute for HAIYUANYU(15), DepartmentofMolecu- GenomicResearch,Rockville,Maryland lar Biophysics and Biochemistry, Yale University,NewHaven,Connecticut JERILYNA.TIMLIN(6),BiomolecularAna- lysis and Imaging, Sandia National La- XIAOWEIZHU(15), Program in Computa- boratories,Albuquerque,NewMexico tionalBiologyandBioinformatics,Yale University,NewHaven,Connecticut CARL TROEIN (7), Computational Biology and Biological Physics, Department of Theoretical Physics, Lund University, Lund,Sweden [1] RNA EXTRACTIONFORARRAYS 1 [1] RNA Extraction for Arrays By LAKSHMI V. MADABUSI, GARY J. LATHAM, and BERNARD F. ANDRUSS Abstract DNA microarrays enable insights into global gene expression by capturing a snapshot of cellular expression levels at the time of sample collection. Careful RNA handling and extraction are required to preserve this information properly, ensure sample‐to‐sample reproducibility, and limit unwanted technical variation in experimental data. This chapter dis- cusses important considerations for ‘‘array‐friendly’’ sample handling and processing from biosamples such as blood, formalin‐fixed, paraffin‐ embedded samples, and fresh or flash‐frozen tissues and cells. It also provides guidelines on RNA quality assessments, which can be used to validate sample preparation and maximize recovery of relevant biological information. Introduction DNA microarrays have enabled biologists to move from the realm of studyingonegeneatatimetounderstandinggenome‐widechangesingene expression. The value of microarray studies has been vetted through nu- merous studies that have linked abnormal transcript levels with many different diseases (Archacki and Wang, 2004; Blalock et al., 2004; Borovecki et al., 2005; Dhanasekaran et al., 2001; Glatt et al., 2005; West et al., 2001). Because these types of studies will be used increasingly to create and validate diagnostic and prognostic expression signatures and tosupporttoxicologicalandfunctionalstudiesthatunderlietheregulatory filings for new drug submissions, it will become increasingly important to create standardized and robust methods for sample procurement, sample processing, anddataanalysis.ThegoalofanyRNAisolationprocedure is to recover an RNA population that faithfully mirrors the biology of the sample at the time of collection. Problems associated with the extraction of biologically representative RNA primarily arise from the susceptibility of RNAtodegradationbyubiquitousandcatalyticallypotentRNases.For tissues and cells, protection of RNA has traditionally been accomplished by immediate lysis using high concentrations of detergents and/or chao- tropic agents and organic solvents (such as TRI reagent). These methods, METHODSINENZYMOLOGY,VOL.411 0076-6879/06$35.00 Copyright2006,ElsevierInc.Allrightsreserved. DOI:10.1016/S0076-6879(06)11001-0 2 DNA microarrays, part B [1] while effective, are complex to use at point of care and suffer from low samplethroughputandpoorstabilizationofcellularRNAforlongperiods. Flashfreezingofthesampleinliquidnitrogenandsubsequenttransporta- tionondryice,althougheffective,areimpracticalinmostclinicalsettings. Finally,diseasespecimenscanpresentbiohazardriskstotheoperatorand constrain sample collection, thus limiting the use of best sample handling and processing practicesand compromising RNA quality. ThepracticalityandefficacyofRNAstabilizationagentssuchasRNA- later to preserve the RNA in tissues, cells, and blood are gaining broad acceptance. Procedures used for collection of samples with RNAlater are simple and can be carried out in a hospital setting with minimal training. Thisreagentisaqueousandnontoxicandallowsconvenienttransportation of samples at ambient temperature. However, RNAlater does not remove the biohazard risks associated with biosamples, and, as a result, all proper safety precautions should be observed. It is beyond the scope of this chaptertoprovidedetailsontherisksassociatedandpreventivemeasures to be taken when dealing with samples considered to be a biohazard. Several regulatory agencies offer guidelines on the safety issues and precautions that need to beaddressed with such samples. In addition to the handling of biological material, limitations can be imposed by the large amounts of RNA necessary for microarray experi- ments.Asaresult,samplessuchastumorbiopsies,formalin‐fixed,paraffin‐ embedded (FFPE)sections, orlasermicrodissectedsamples requireRNA amplificationtogenerateadequateamountsoflabeledmaterialformicro- array hybridization. The most popular and best validated approaches for amplifying RNA are based on the linear RNA amplification method de- veloped byEberwine(VanGelder,1990).Thistechniquehasbeenwidely acceptedformicroarrayapplicationsandisknowntopreservetheoriginal transcriptratiosinthesample(Feldmanetal.,2002;Polaceketal.,2003).In terms of RNA quality, parameters such as A260:280 measurements and Agilent RNA integrity number (RIN) are often used to gauge the quality of samples and predict their suitability for microarray studies. The mini- mum A260:280 or RIN number suitable for analysis varies by the array platform, number of replicates, and the experimental questions to be answered inthe study. Blood as a Biological Specimen Blood is a highly desirable biosample for research and clinical studies for several reasons. First, blood is highly accessible and can be collected using relatively simple methods. Second, limited infrastructure is required to draw blood from a large number of patients. Finally, blood circulates