1. The medical relevance of genome variability Gabor T. Marth, D.Sc. Department of Biology, Boston College marth@bc.edu

2. Lecture overview 1. Phenotypic effects caused by known genetic variants 2. Genetic mapping to find genetic variants that cause diseases – linkage analysis and association studies 3. Genome-wide association mapping resources – the HapMap 4. Structural and epigenetic variations in disease

3. 1. Phenotypic effects caused by known genetic variants

4. Many SNPs do have phenotypic effects Badano and Katsanis, NRG 2002 some notable genetic diseases: cystic fibrosis cycle-cell anemia

5. Genetic variants in Pharmacogenetics Evans and Rellig, Science 1999

6. Genetic variants in Pharmacogenetics Evans and Rellig, Science 1999

7. Using genotype information in the drug development pipeline Roses. NRG 2004

8. Are all genetic variants functional? ~ 10 million known SNPs SNPs, on the scale of the genome, can be described well with the “neutral theory” of sequence variations the vast majority of SNPs likely to have no functional effects How do we find the few functional variants in the background of millions of non-functional SNPs?

9. 2. Genetic mapping to find genetic variants that cause diseases – linkage analysis and association studies

10. Genetic mapping

11. Allelic association (linkage) allelic association is the non- random assortment between alleles i.e. it measures how well knowledge of the allele state at one site permits prediction at another marker site functional site significant allelic association between a marker and a functional site permits localization (mapping) even without having the functional site in our collection allelic association, and the use of genetic markers is the basis for mapping functional alleles

12. Mendelian diseases have simple inheritance genotype inheritance genotype + phenotype inheritance

13. Linkage analysis compares the transmission of marker genotype and phenotype in families

14. Complex disease – complex inheritance Badano and Katsanis, NRG 2002

15. Allele frequency and relative risk Brinkman et al. Nature Reviews Genetics advance online publication; published online 14 March 2006 | doi:10.1038/nrg1828

16. Association study strategies region(s) interrogated: single gene, list of candidate genes (“candidate gene study”), or entire genome (“genome scan”) direct or indirect: causative variant marker that is co-inherited with causative variant single-SNP marker or multi- SNP haplotype marker single-stage or multi-stage

17. Association study strategies 2. LD-driven – based entirely on the reduction of redundancy presented by the linkage disequilibrium (LD) between SNPs; tags represent other SNPs they are correlated with 1. hypothesis driven (i.e. based on gene function) causative variant for economy, one cannot genotype every SNP in thousands of clinical samples: marker selection is the process where a subset of all available SNPs is chosen

18. Marker selection depends on genome LD Daly et al. NG 2001

19. Case-control association testing searching for markers with “significant” marker allele frequency differences between cases and controls; these marker signify regions of possible causative alleles AF(cases) AF(controls) clinical cases clinical controls genotyping cases and controls at various polymorphisms

20. 3. Genome-wide association mapping resources – the HapMap

21. The HapMap resource goal: to map out human allele and association structure of at the kilobase scale deliverables: a set of physical and informational reagents

22. LD structure in four human populations International HapMap Consortium, Nature 2005

23. LD varies across samples African reference (YRI) there are large differences in LD between different human populations… European reference (CEU) … and even between samples from the same population. Other European samples

24. Sample-to-sample LD differences make tagSNP selection problematic groups of SNPs that are in LD in the HapMap reference samples may not be in a future set of clinical samples… … and tags that were selected based on LD in the HapMap may no longer work (i.e. represent the SNPs they were supposed to) in the clinical samples… … possibly resulting in missed disease associations.

25. Marker selection with additional samples test if markers selected from the HapMap continue to “tag” other SNPs in their original LD group

26. Representative computational samples

27. Two methods of computational sample generation “HapMap” “cases” “controls” HapMap Method 1. “Data-relevant Coalescent”. This algorithm uses a population genetic model to connect mutations in the HapMap reference to mutations in future clinical samples. Full model but computationally slow. Method 2. The PAC method (product of approximate conditionals, Li & Stephens). This method constructs “new” samples as mosaics of existing haplotypes, mimicking the effects of recombination. An approximation but fast.

28. LD difference -- comparison to extra experimental genotypes 0.949 +/- 0.013 0.978 +/- 0.010 0.963 +/- 0.014 we have analyzed two extra genotype sets collected at the HapMap SNPs in three genome regions, from our clinical collaborators (Prof. Thomas Hudson, McGill; Prof. Stanley Nelson, UCLA)

29. Genome-wide scans for human diseases Klein et al, Science 2005 SNPs in Complement Factor H (CFH) gene are associated with Age-related Macular Degeneration (AMD)

30. 4. Somatic, structural and epigenetic variants in disease

31. Somatic mutations © Brian Stavely, Memorial University of Newfoundland the detection of somatic mutations, and their distinction from inherited polymorphism, is important to separate pre-disposing variants from mutations that occur during disease progression e.g. in cancer 1. detect the mutations 2. classify whether somatic or inherited

32. Detecting somatic mutations with comparative data based on comparison of cancer and normal tissue from the same individual often cancer tissue is highly heterogeneous and the somatic mutant allele may represent at low allele frequency

33. Detecting somatic mutations with subtraction if normal tissue samples are not available, we detect SNPs in cancer tissue against e.g. the human genome reference sequence subtract apparent mutations that are present in sequence variation databases search for evidence that these mutations are genetic

34. Detecting somatic mutations in murine mtDNA we have applied our methods for somatic mutation detection in murine mitochondrial sequences heteroplasmyhomoplasmy we will be applying our methods for human nuclear DNA from our collaborators

35. Structural variants in disease Feuk et al. Nature Reviews Genetics 7, 85–97 (February 2006) | doi:10.1038/nrg1767

36. Structural variations and phenotype Feuk et al. Nature Reviews Genetics 7, 85–97 (February 2006) | doi:10.1038/nrg1767

37. Epigenetics and cancer Baylin at al. NRC 2006.

38. Informatics of detection / integration of varied genetic and epigenetic data chromatin structure gene expression profiles copy number changes methylation profiles chromosome rearrangements repeat expansions somatic mutations