--- module: 092-02 language: en chapter: 92 title: "Chromosomes, Human Variation, Population Genetics, and Genomic Methods" module_title: "Human variation, penetrance, complex traits, population genetics, and association" source_sha256: 43bd3ef9774b4a587aee9129d280914601aaa31e901a6a91e1db1ce19054a2e5 --- # Human variation, population genetics, and association ## Variant classes and frequency ### Millions of variants, mostly neutral or small effect ### Separate ancestry from race, probability from destiny ### Single nucleotides, indels, copy number, structure ### Short tandem repeats may expand unstably ### Frequency counted among chromosomes, not people ### Rare is not pathogenic, common not harmless #### Recessive, late-onset alleles can become frequent ## Mutation, drift, selection ### Mutation rates vary by context and parental age #### CpG cytosines mutate after deamination #### Recurrent mutation in unrelated families ### Drift strongest in small populations #### Founder effect and bottleneck raise rare alleles #### Founder testing must not replace broad analysis ### Negative selection removes harmful alleles ### Balancing selection maintains variation #### Sickle heterozygotes: less severe malaria ## Ancestry, race, gene flow ### Migration moves alleles between populations ### Variation shared, no discrete biological races ### Genetic ancestry is statistical similarity ### Race reflects exposures, not genotype ## Hardy–Weinberg equilibrium ### Random mating, large population, no disturbing forces ### p squared, 2pq, q squared ### A null model, not a claim about populations ### Departure: structure, inbreeding, selection, error ### Carrier frequency about twice allele frequency #### Fails with founders, assortative mating #### Use measured population data ## Linkage and linkage disequilibrium ### Nearby loci transmitted together ### Recombination fraction at most one-half ### Centimorgans differ from base pairs #### Vary by sex and region ### Hotspots shape haplotype blocks ### Linkage disequilibrium: nonrandom allele association #### Marker tags an unmeasured causal variant #### Strength and direction differ by ancestry #### Fine mapping seeks the causal set ## Heritability and complex traits ### Heritability: variance share in a context #### Not an individual's genetic share #### High heritability yet diet treats phenylketonuria ### Broad includes all, narrow is additive ### Twin studies rest on assumptions ### Common variants small, rare variants larger effects ### Liability threshold for binary disease ### Familial clustering need not be Mendelian ### Interaction depends on statistical scale ### Gene–environment correlation mimics interaction ## Genome-wide association ### Quality control of samples and genotypes ### Stringent significance for multiple testing ### Association alone does not establish mechanism ### Stratification confounds association #### Principal components reduce, not eliminate #### Diverse recruitment improves transferability ### Odds ratio is not absolute risk #### Overstates risk ratio when outcome common ### Winner's curse inflates discovery effects ## Scores, instruments, rare variants ### Polygenic score sums weighted risk alleles #### Stratifies relative risk within populations #### High is not diagnosis, low is not exclusion ### Mendelian randomisation: variants as instruments #### Pleiotropy and structure violate assumptions ### Rare variants aggregated by gene or region #### Burden assumes same direction, kernel mixed ## Bias and interpretation limits ### Family penetrance inflated by ascertainment ### Biobanks underrepresent severe childhood disease ### Collider bias from selective participation ### Records misclassify missing diagnoses as absence ### Large samples cannot repair systematic bias ### Database absence supports rarity only ### Three levels: molecule, probability, population