2007. stephen chanock. technologic issues in gwas and follow up studies
- 1. Technologic Issues in GWAS and Follow-up Studies Stephen Chanock, M.D. Senior Investigator, POB,CCR & Director, Core Genotyping Facility, DCEG NCI May 22, 2007
- 2. Types of Polymorphisms Single nucleotide polymorphisms (SNPs) Common SNPs are defined as >1% in at least one population Rare SNPs are hard to identify and validate But, it is estimated that there are a large number per individual MAF= minor allele frequency
- 3. SNP in Unique Sequence T/C
- 4. SNPs & Function: We know so little.. • Majority are “silent” – No known functional change • Alter gene expression/regulation – Promoter/enhancer/silencer – mRNA stability – Small RNAs • Alter function of gene product – Change sequence of protein
- 5. SNPs in Genes: Take one Coding SNPs Synonymous- no change in amino acid previously termed “silent” but….. Can alter mRNA stability DRD2 (Duan et al 2002) Nonsynonymous- changes amino acid conservative and radical Nonsense- insertion of stop codon Indel- Disrupts codon sequence Rare but disruptive
- 6. SNPs Outside Genes: Take many…. • Majority distributed throughout genome are “silent” (excellent as markers) • Alter transcription – Promoter, enhancer, silencer • Regulate expression – Locus control region, mRNA stability • Most are assumed to be ‘silent hitchhikers’ – No function by predictive models or analysis
- 7. Linkage disequilibrium (LD) • The non-random association of alleles in the population • Alleles at neighboring loci tend to cosegregate • Linkage disequilibrium implies population allelic association
- 8. Strategy for SNP Selection To test all SNPs is presently too costly Utilize a strategy that capitalizes on linkage disequilibrium between SNPs Haplotype blocks defined by Gabriel et al Based on D’ values for linkage disequilibrium
- 9. What can LD do for me? • Knowledge of patterns of linkage disequilibrium can be quite useful in the design and analysis of genetic data • Design: – Estimation of theoretical power to detect associations – Evaluation of degree completeness of sampling of genetic variants – Choice of most informative genetic variants to genotype
- 10. Genetic Association Testing: Finding Markers
- 11. Large and Small Scale Polymorphisms • Copy Number of Polymorphisms Regional “repeat” of sequence 10s to 100s kb of sequence Estimate of >10% of human genome Multi-copy in many individuals • Duplicons 90-100% similarity for >1 kb 5-10% of genome (5% exons elsewhere) Multi-copy (high N) in all individuals
- 12. Germ-line DNA Copy Number Variation(CNV)
- 13. Copy Number Variation Across the Genome Redon et al Nature 2006
- 14. Copy Number Variation Across the Genome Frequency of CNVs Most are uncommon (<5%) Familial vs Unrelated Studies Associated with Disease OPN1LW Red/green colorblind CCL3L1 Reduced HIV Infection CYP2A6 Altered nicotine metabolism VKORC1 Warfarin metabolism
- 15. Most SNPs Are In Unique Sequence ! T/C
- 16. PSV (Paralogous Sequence Variant) A G
- 17. AA/C SNP in Duplicon Sequence
- 18. MSV: Multi-Site Variants T G GT G T G T/G T/G T/G
- 19. Progress in Genotyping Technology 1 10 102 103 104 105 106 # of SNPs Cost per genotype Cents (USD) 10 1 102 ABI TaqMan ABI SNPlex Illumina Golden Gate Illumina Infinium/Sentrix Affymetrix 100K/500K Perlegen Affymetrix MegAllele 2001 2007 Affymetrix 10K Sequenom
- 20. Genotype Technologies • Dropping costs • Smaller amounts of DNA < 1ug for > 1 Million SNPs • Economy of scale Frequent Flier Paradigm • Increased density but with fixed products • Custom products bear high development cost • Challenge of mid-range (50 to 500 SNPs)
- 21. from http://docs.appliedbiosystems.com/pebiodocs/00790910.pdf TaqManTM (5’ Exonuclease)
- 22. BRCA1-03 rs16941 SNP under both TM probes Results in the wrong allele call SNP Analysis- BEWARE http://snp500cancer.nci.nih.gov
- 23. Infinium™ Assay Address 1 Address 2 Address 3 Array of capture probes (1 bead type/allele) Address 4 Address 1 Address 2 Address 3 Extension and uniform, single-color labeling Address 4 Amplify genome uniformly Fragment, Denature and Hybridize to immobilized 50- mers Discriminate SNPs (enzyme-mediated) Amplify signal and readout 1 3/4 “A/A” “A/B” 2 1 2 3 4
- 24. Illumina HumanHap500 Good Cluster Poor Cluster Read in BeadStudioTM A/A A/G G/G ?/? ?/?* ?*/?*
- 25. 250 ng Genomic DNA RE Digestion Adaptor Ligation Affymetrix GeneChip® Mapping Assay Xba XbaXba Fragmentation and Labeling PCR: Universal Primer Amplification Complexity Reduction AA BB AB Hyb & Wash
- 26. Affymetrix 500K Chips Poor Quality Good Quality
- 27. Important Points • Too many data points to review individually • Iterative algorithm for analysis – Still undergoing improvements • Validation of notable SNPs with second technology – “Neighboring SNP-land mines” • Do not do this at home- Only for highly trained personnel
- 28. Choice of Dense SNP Platforms Basic Points Based on ‘Spacing’ 100k, 500k, 1 Million CNV Analysis WGA compatible Issues Lower price 2 Enzyme Problem Calling Algorithms Redundancy (useful) Basic Points Based on ‘tagging’ 317k, 550k, 1 Million CNV Analysis WGA not yet rec’d Issues Higher price HapMap II Based Affymetrix Illumina
- 29. 2007 What is Available for Whole Genome Scans • Coverage analysis based on HapMap II Data • Build 20 MAF >5%, r2 > 0.8 (pair-wise) • CEU YRI JPT/CHB • Illumina HumanHap300 80% 35% 40% • Illumina HumanHap500 91% 58% 88% • Affymetrix* 500k Mapping 63*% 41% 63% *77% (with 50k MegA)
- 30. Quality control of genotype calls & DNA handling
- 31. Quality Control for Called Genotypes PURPOSE: Identification of unreliable SNPs and DNAs to be entirely removed from the analysis . Evaluation of completion rate (DNAs) Evaluation of call rate (SNPs) Evaluation of discordance rate (error rate)
- 32. Attempted SNPs CGEMS SNP DNA Success Rates Threshold DNA completion rate > 94% attempted failed Success rate 4696 66 0.986 Criterium SNP call rate > 90% attempted failed success rate PLCO 561,494 1,490 0.973 NHS 555,352 8,706 0.984 Number of individuals Number of individuals 1.4% 2.1% AttemptedDNAs SNPx DNA y
- 33. Discordance rate for CGEMS: HumanHap500 (Illumina) CEPH-CGEMS 74 duplicate pairs Mean discordance rate 2 x 10-4 Participants 142 duplicate pairs Mean discordance rate Prostate 2.0 x 10-4 Breast 1.5 x 10-4 CEPH- HapMap 28 individuals (with 24 duplicates) Mean discordance rate 1.4 x 10-3 Concordance rates >99.5% Subtle Differences in Quality of DNA
- 34. Quality Control for Recruitment DNA handling Checking for: Chromosome X Ploidy Identification of Familial Relationships Evaluation of Continental Admixture Population Stratification Principal Component Analysis (Hardy Weinberg Statistics)
- 35. Chromosome X 1 copy 2 copies Prostate 2279 3 Breast 0 Unexpected duplicates 1st & 2nd degree relatives Other 3rd to 5th degree relatives 3 pairs 5 20* *It was noted subsequently that both members of each pair had been recruited in the same center. 1 to be done3 pairs2299 Analysis of CGEMS Data Sets
- 36. Admixture coefficient in PLCO samples Asia AfricaEurope control case
- 37. -.15 -.1-.05 0.05 V_Dim1 -.05 0 .05 .1 V_Dim2 33 controls 21 cases 69 controls 91 cases CGEMS Prostate Cases & Controls Principal Component Analysis Based on Price et al Nat Genet 2006
- 38. 0 log10(p-value) log10(quantile) corrected uncorrected -2 -4 -6 0-2-4-6 CGEMS Breast Cancer Scan log quantile plot of p-values for the Entire Set of Markers
- 39. Chromosomes 1 2 3 4 5 6 7 8 -2 -4 -6 -4 -6 Log10(p-value) p q p q 222120191817161514131211109 X -2 p q p q incidence density sampling 8q24 CGEMS Prostate Cancer GWAS P values < 0.01
- 40. Replication Studies in CGEMS Prostate Cancer GWAS PLCO ACS ATBC FPCC HPFS ALL Subjects 1157 1172 Predisposing allele frequency 0.55 0.49 1151 1150 0.55 0.50 896 894 0.57 0.51 459 455 0.56 0.51 636 625 0.57 0.51 4299 4296 0.56 0.50 P-value 2.4x10-05 3.2x10-03 1.9x10-03 1.2x10-01 1.0x10-02 9.4x10-13 Predisposing allele frequency 0.14 0.10 0.12 0.08 0.21 0.17 0.12 0.07 0.13 0.09 0.15 0.11 P-value 9.8x10-05 2.7x10-05 2.9x10-02 4.4x10-03 2.7x10-03 1.5x10-14 rs6983267 rs1447295 Cases Cont. Cases Cont. Estimated Odds Ratios Overall Heterozygotes 1.26 1.43 Homozygotes 1.58 2.23
- 41. Meta-Analysis of 8q24 papers in Nature Genetics: J Witte J Witte Nature Genetics 2007