Polygenic Risk Scores for Medical Research PDF

Genomewide association studies and polygenic risk scores. Gerome Breen @psychgenomics Professor of Psychiatric Genetics Social, Genetic, and Developmental Psychiatry Centre...

Genomewide association studies and polygenic risk scores. Gerome Breen @psychgenomics Professor of Psychiatric Genetics Social, Genetic, and Developmental Psychiatry Centre Institute of Psychiatry, Psychology, and Neuroscience. King’s College London  Lecture Outline Background: disease genes, genetic association testing and inheritance models Genome-Wide Association Studies (GWAS): what they are and how they work From discovery to prediction in genetics Polygenic Risk Scores for Medical Research  Mendelian genetic traits Polygenic disorders Huntington’s disease, Cystic fibrosis, Alpha-1 Asthma, arthritis, cancer, coronary heart disease, antitrypsin deficiency … Zellweger syndrome dementia, depression, diabetes, psychosis, stroke Single gene Multiple genes Sequencing (expensive) Genotyping (cost-effective) 100,000 Genomes project, DDD, … Few (but increasing) funded initiatives Moving to routine clinical care Needs stronger research base for Rare disorders, rare variants translation Individual impact: high Common disorders, common variants Population impact: low Individual impact: low to moderate Population impact: high The ‘gene for’ news stories.. The ‘gene for’ news stories… There are in fact usually many genes that affect any complex outcome The ‘gene for’ news stories… Each 'gene' usually has a tiny effect (ie. they do not cause the disease by themselves)  The ‘gene for’ news stories… These things are not genes! They are genetic variations – changes in the DNA code.  Genetic Association Testing Now we know that most diseases and human traits are caused by many genetic variants. These are mostly of tiny effect, but although collectively mighty. But how are these genetic risk variants found?  Association testing in genetic research Different alleles at a SNP can also have an effect on diseases/disorders. Genetic Association Studies test the association between the alleles/genotypes of a SNP and a trait of interest  Genotype Inheritance Models The pattern of effects that the different genotypes have on the phenotype are categorised into different inheritance models. Above are illustrations of two main inheritance models shown in relation to a continuous trait - height. For common variants Additive effects are overwhelming what is found  Genome wide association: GWAS Goal: Uncover the genetic basis of a given disease Basic Idea: examination a significant proportion of all common genetic variation across the human genome, in order to to identify genetic associations Look for associations between genotypes at each locus and disease status. Multiple technical challenges were overcome. 12  Linkage Disequilibrium Non-random assortment of alleles at 2 or more loci (~correlation) The closer the markers, the stronger the LD since recombination will have occurred at a low rate Markers in strong LD co-segregate families and within the population  SNP selection for genome wide studies Choose common SNPs Representative tag set accounting for linkage disequilibrium (how genetic variants correlate with each other) Capture most common variation  Imputation Once you have a GWAS of tagging SNPs we run the process in reverse – “imputation”. Typically a GWAS increases from 500K-1M genotyped SNP markers to 6-12M imputed+genotyped genetic variants TA Manolio et al. Nature 461, 747-753 (2009) doi:10.1038/nature08494 GWAS Discovery vs Sample Size Height" 100" Crohn's"disease" Body"mass"index" Prostate"cancer" 100" QT"interval" Breast"cancer" HDL"cholesterol" Type"2"diabetes" Bone"mineral"density" Ulcera:ve"coli:s" #"GWAS"hits" #"GWAS"hits" 10" 10" 1" 1" 2000" 20000" 200000" 1000" 10000" 100000" Discovery"sample"size" #"cases" Visscher et al. 2012 Am J Hum Genet 17  Lecture Outline I Background: gene finding, genetic association testing and inheritance models I Genome-Wide Association Studies (GWAS): what they are and how they work I From discovery to prediction in genetics I Polygenic Risk Scores for Medical Research  Beyond GWAS: Personalised Medicine IA major ambition of the genomic era is to use genetics to predict what diseases and disorders different people may get later in life I This could initiate personalised/precision medicine I Wecan use genetics to predict human traits such as height, blood pressure and cholesterol levels, as well as IQ, alcohol consumption and extraversion - so it is not only disease From discovery toprediction From discovery toprediction From discovery toprediction From discovery toprediction From discovery toprediction Genetic Prediction - doomed?  Genetic Prediction - doomed? Individual-level disease prediction from genetics is still very challenging (but is changing now for diseases such as breast cancer and cardiovascular disease) However, in all disorders/diseases aggregating predictions across samples allows powerful group-level inference and many different applications  Prediction in groups of people Allen et al. 2010 Genome-wide significant SNPs for BMI are predictive of BMI Why predict in groups ofpeople? Why predict in groups ofpeople? Why predict in groups ofpeople?  Why predict in groups ofpeople? There are a huge number of scientific questions that can be answered by comparing predictions in groups of people With larger GWAS samples, predictions at the individual-level is becoming possible for an increasing number of traits  How can wedo prediction from genetics? The primary aim of GWAS is susceptibility locus DISCOVERY with variants declared only if we are certain (p