Rare and common variants: twenty arguments G.Gibson Homework 3 Mylène Champs Marine Flechet Mathieu Stifkens Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 1 Content : Rare and common variants Introduction Summary ◦ Rare allele model ◦ Infinitesimal model Conclusion Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 2 Content : Rare and common variants Introduction Summary ◦ Rare allele model ◦ Infinitesimal model Conclusion Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 3 Introduction: Rare and common variants ◦ Genome-wide association studies (GWASs) identify genetic factors that influence health and disease. ◦ First model used : CDCV (Common disease Common variant) = a small number of common variants can explain the percentage of disease risk. ◦ This model is not used anymore because of the “missing heritability problem”. A few loci with moderate effect cannot explain several percent of disease susceptibility. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 4 Content : Rare and common variants Introduction Summary ◦ Rare allele model ◦ Infinitesimal model Conclusion Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 5 Summary : Rare and common variants Rare allele model ◦ Presentation of the model ◦ Arguments « in favour » ◦ Arguments « against » Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 6 Summary : Rare and common variants Rare allele model – Presentation ◦ Model known as « many rare alleles of large effect ». ◦ The variance for a disease is due to rare variants (allele frequency<1%) which are highly penetrant (large effect). ◦ Example: Schizophrenia = collection of many similar conditions that are attributable to rare variants. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 7 Summary : Rare and common variants Rare allele model – Presentation Causal variant effects (yellow dots) may be large in a few individuals but are not common enough to represent a “hit” in a GWAS. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 8 Summary : Rare and common variants Rare allele model ◦ Presentation of the model ◦ Arguments « in favour » ◦ Arguments « against » Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 9 Summary : Rare and common variants Rare allele model – « In favour » ◦ Evolutionnary theory predicts that disease alleles should be rare[1] ; ◦ Empirical population genetic data shows that deleterious variants are rare[1] ; ◦ Rare copy number variants contribute to several complex psychological disorders[1] ; ◦ Many rare familial disorders are due to rare alleles of large effects[1]; ◦ Synthetic association may explain common variants effects[1] . Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 10 Summary : Rare and common variants Rare allele model – « In favour » ◦ Evolutionnary theory predicts that disease alleles should be rare[1] ; ◦ Empirical population genetic data shows that deleterious variants are rare[1] ; ◦ Rare copy number variants contribute to several complex psychological disorders[1] ; ◦ Many rare familial disorders are due to rare alleles of large effects[1]; ◦ Synthetic association may explain common variants effects[1] . Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 11 Summary : Rare and common variants Evolutionnary theory predicts that disease alleles should be rare[1] : ◦ Deleterious alleles are created by mutation; removed by purifying selection. ◦ Rate(creation) > rate (removal) Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 12 Summary : Rare and common variants Rare copy number variants contribute to several complex psychological disorders[1] : ◦ CNVs : hemizygous deletion – local duplication; ◦ Promote disease such as schyzophrenia and autism and modify its severity . Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 13 Summary : Rare and common variants Synthetic association may explain common variants effects[1] : LD Data [2] For common variants which do not explain much percentage of the disease susceptibility Rare variants increase this case risk. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 14 Summary : Rare and common variants Rare allele model ◦ Presentation of the model ◦ Arguments « in favour » ◦ Arguments « against » Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 15 Summary : Rare and common variants Rare allele model – « Against » ◦ Analysis of GWAS data is not consistent with rare variants explanations[1] ; ◦ Sibling recurrence rates are greater than would be expected by the postulated effect sizes of rare variants[1] ; ◦ Rare variants do not obviously have additive effects[1] ; ◦ Epidemiological transitions cannot be attributed to rare variants[1] ; ◦ GWAS associations are consistent across populations[1] ; Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 16 Summary : Rare and common variants Rare allele model – « Against » ◦ Analysis of GWAS data is not consistent with rare variants explanations[1] ; ◦ Sibling recurrence rates are greater than would be expected by the postulated effect sizes of rare variants[1] ; ◦ Rare variants do not obviously have additive effects[1] ; ◦ Epidemiological transitions cannot be attributed to rare variants[1] ; ◦ GWAS associations are consistent across populations[1] ; Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 17 Summary : Rare and common variants Analysis of GWAS data is not consistent with rare variants explanations[1] ◦ Rare variants cannot be the predominant source of heritabilily; ◦ There should be many of them with large size and effect. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 18 Summary : Rare and common variants Rare variants do not obviously have additive effects[1] ◦ Genetic associations are known to be additive whereas rare variants interact multiplicatively and they have dominant effect; ◦ However on the statistical side rare variants induce additivity effects. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 19 Summary : Rare and common variants Epidemiological transitions cannot be attributed to rare variants[1] ◦ The change of prevalence of some diseases is too fast; ◦ The model can not explain the influence of environmental variable. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 20 Content : Rare and common variants Introduction Summary ◦ Rare allele model ◦ Infinitesimal model Conclusion Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 21 Summary : Rare and common variants Infinitesimal model ◦ Presentation of the model ◦ Arguments « in favour » ◦ Arguments « against » Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 22 Summary : Rare and common variants Infinitesimal model – Presentation ◦ Known as « common » model or many common variants of small effects. ◦ This is the model used in GWASs. ◦ Common variants are the major source of genetic variance for disease susceptibility. ◦ Hundreds or thousands of loci of small effect contribute in each case. ◦ Example : Height or BMI studies, hundred of loci have been found but they don’t explain all of the genetic variance. This problem is called the « missing heritability problem ». Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 23 Summary : Rare and common variants Infinitesimal model – Presentation Significant “hits” of common variants with small effects. Several SNPs within a linkage disequilibrium (LD) block are associated with the trait [1]. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 24 Summary : Rare and common variants Infinitesimal model ◦ Presentation of the model ◦ Arguments « in favour » ◦ Arguments « against » Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 25 Summary : Rare and common variants Infinitesimal model – « In favour » ◦ The infinitesimal model underpins standard quantitative genetic theory[1] ; ◦ Common variants collectively capture the majority of the genetic variance in GWASs[1] ; ◦ Variation in endophenotypes is almost certainly due to common variants[1] ; ◦ Model organism research supports common variants contributions to complex phenotypes[1] ; ◦ GWASs have successfully identified thousands of common variants[1] . Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 26 Summary : Rare and common variants Infinitesimal model – « In favour » ◦ The infinitesimal model underpins standard quantitative genetic theory[1] ; ◦ Common variants collectively capture the majority of the genetic variance in GWASs[1] ; ◦ Variation in endophenotypes is almost certainly due to common variants[1] ; ◦ Model organism research supports common variants contributions to complex phenotypes[1] ; ◦ GWASs have successfully identified thousands of common variants[1] . Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 27 Summary : Rare and common variants The infinitesimal model underpins standard quantitative genetic theory[1] : ◦ High heritability ; ◦ No results were against the infinitesimal model. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 28 Summary : Rare and common variants Common variants collectively capture the majority of the genetic variance in GWASs[1]: Capture more of the genetic variance by using all significant SNPs; Variance is attributed to hundreds of loci. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 29 Summary : Rare and common variants GWASs have successfully identified thousands of common variants[1] : ◦ Unrealistic assumptions of the effect size ; ◦ Increasing samples allows to determine more loci. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 30 Summary : Rare and common variants Infinitesimal model ◦ Presentation of the model ◦ Arguments « in favour » ◦ Arguments « against » Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 31 Summary : Rare and common variants Infinitesimal model – « Against » ◦ The QTL paradox[1] ; ◦ The abscence of blending inheritence[1] ; ◦ Demographic phenomena suggest more than one simple common-variant model[1] ; ◦ Very few common variants for disease have been functionnaly validated[1] ; ◦ What accounts for the missing heritability[1] ? Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 32 Summary : Rare and common variants Infinitesimal model – « Against » ◦ The QTL paradox[1] ; ◦ The abscence of blending inheritence[1] ; ◦ Demographic phenomena suggest more than one simple common-variant model[1] ; ◦ Very few common variants for disease have been functionnaly validated[1] ; ◦ What accounts for the missing heritability[1] ? Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 33 Summary : Rare and common variants The QTL paradox[1] ◦ We cannot find QTLs detected in pedigrees and in experimental crosses; ◦ Explanations: -> QTLs are rare variants that only contribute in that cross. -> Each cross captures only a small fraction of genetic variance in a population. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 34 Summary : Rare and common variants The abscence of blending inheritence[1] ◦ The granularity in the distribution of risks and phenotypic trait variation should decrease with the crossing of two unrelated poeple; ◦ However we observe higher risks than the model predicted; ◦ For example : We can observe that in some family complex phenotype traits are recurrent. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 35 Summary : Rare and common variants What accounts for the missing heritability[1] ? ◦ The model does not take into account the missing heritability problem; ◦ But the problem really exists ! Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 36 Content : Rare and common variants Introduction Summary ◦ Rare allele model ◦ Infinitesimal model Conclusion Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 37 Conclusion : Rare and common variants Which model would you choose ? Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 38 Conclusion : Rare and common variants Which model would you choose ? ◦ Both ! ◦ We should learn how to use the two models together because they both have their place in the current research. ◦ Idea : Integrate rare and common variants effects together. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 39 Conclusion : Rare and common variants The common variants establish the background liability to a disease and this liability can be modified by the rare variants with large effects [1]. Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 40 Thank you for your attention ! Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 41 References : [1] G. GIBSON : Rare and common variants: twenty arguments. Nat. Rev. Genet., 13(2):135145, Feb 2012. [2] Bioinformatics course – GWAS studies, K. VAN STEEN Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 42 Do you have any question(s) ? Bioinformatics - GBIO0009-1 - K.Van Steen University of Liège 43