1. Large-scale recombination rate patterns are conserved among human populations David Serre McGill University and Genome Quebec Innovation Center UQAM January 2006

2. - Recombination? - Why is recombination important? - How do we measure recombination rates? - Large scale recombination rate differences among human populations Recombination rate patterns in Humans

5. (for gene conversion see Padhukasahasram AJHG 2004)

13. Why Studying Recombination ? Recombination separates history of two loci located on same chromosome.

14. n+1 generation Recombination and disease mapping

15. n

16. n-1 Recombination and disease mapping

17. n-1 Recombination and disease mapping

18. n-2 Recombination and disease mapping

19. n-20 Recombination and disease mapping

20. today Recombination and disease mapping

21. today Recombination and disease mapping

22. Neutral locusLocus under selection Hitchhiking effectLow recombination rate Independent evolutionHigh recombination rate Sabeti et al, 2002 Recombination and natural selection

23. Recombination DirectIndirect Genetic map Sperm typing Linkage Disequilibrium Population history Understanding Recombination…

24. How often do we observe a recombination event between two markers ? Divide recombination fraction by physical distance (bp)  recombination rate - Marshfield map Yu et al., 2001 - DeCode map Kong et al., 2002 Kong et al., 2004 Genetic maps

25. ba BA How often do we observe a recombinant spermatozoid ? Divide recombination fraction by physical distance (bp)  recombination rate Papers from Alec Jeffreys and Norm Arnheim Sperm typing

26. “non-random association of alleles” Aa Bp(A)p(B)(1-p(A))p(B) bp(A)(1-p(B))(1-p(A))(1-p(B)) If Pr(AB) ≠ P(A)*P(B) Then markers are in LD Random-association  no linkage  linkage “equilibrium” Non-random association  linkage  linkage disequilibrium Linkage Disequilibrium (LD)

27. If Pr(AB) ≠ P(A)*P(B) then markers are in LD |D’|: D’= p(AB) – p(A)p(B) influenced by allele frequency R 2 : R 2 =(p(AB) – p(A)p(B)) 2 / (p(A)p(B)p(a)p(b)) normalized chi-squared distributed Measures of LD

28. Dawson et al, 2002

30. Polymorphism data Measure LD (Given a model of population history) Estimate recombination: ρ=4N e c Inferring Recombination from LD

31. Pritchard and Przeworski, 2001 neutralgrowthstructure

32. LD reflects the number/location of historical recombination events. For a given population size, A given population history, What is the most likely recombination rate that would produce the observed LD pattern? Hudson 2001, McVean 2004 From LD to Recombination rates

33. LD betweenLikelihood | ρ=0.001Likelihood | ρ=0.002…Likelihood | ρ=1.000 SNP1-SNP2 SNP1-SNP3 SNP1-SNP4 SNP1-SNP5 SNP2-SNP3 SNP2-SNP4 SNP2-SNP5 SNP3-SNP4 … Total

34. -Fine-scale variations (<100kb) -Individual variation in hotspot intensity (sperm typing) (Jeffreys 1998, 2002, 2005) -Inter-species variation in hotspot intensity and/or location (LD) (Wall 2003, Ptak 2004, 2005, Winckler 2005) -Population differences in hotspot location and/or intensity ? (Crawford 2004) Recombination variations in Humans

35. -Fine-scale variations (<100kb) -Individual variation in hotspot intensity (sperm typing) (Jeffreys 1998, 2002, 2005) -Inter-species variation in hotspot intensity and/or location (LD) (Wall 2003, Ptak 2004, 2005, Winckler 2005) -Population differences in hotspot location and/or intensity ? (Crawford 2004) Recombination variations in Humans

36. -Fine-scale variations (<100kb) -Individual variation in hotspot intensity (sperm typing) (Jeffreys 1998, 2002, 2005) -Inter-species variation in hotspot intensity and/or location (LD) (Wall 2003, Ptak 2004, 2005, Winckler 2005) -Population differences in hotspot location and/or intensity ? (Crawford 2004) Recombination variations in Humans

37. -Large-scale variations (>1Mb) -Individual variation in number of recombination events per meiosis (Yu 1996, Broman 1998, Kong 2002, 2004) -Inter-species variation in genetic map length (Rogers 2000) Recombination variations in Humans

38. -Large-scale variations (>1Mb) -Individual variation in number of recombination events per meiosis (Yu 1996, Broman 1998, Kong 2002, 2004) -Inter-species variation in genetic map length (Rogers 2000) Recombination variations in Humans “The total centimorgan distances among homologous markers are 28.0% longer in the human genome than in the baboon, suggesting that rates of recombination may be higher in humans.” (Rogers et al. 2000)

39. -Fine-scale variations (<100kb) -Individual variation in hotspot intensity (sperm typing) (Jeffreys 1998, 2002, 2005) -Inter-species variation in hotspot intensity and/or location (LD) (Wall 2003, Ptak 2004, 2005, Winckler 2005) -Population differences in hotspot location and/or intensity ? (Crawford 2004) -Large-scale variations (>1Mb) -Individual variation in number of recombination events per meiosis (Yu 1996, Broman 1998, Kong 2002, 2004) -Inter-species variation in genetic map length (Rogers 2000) Recombination variations in Humans

40. -Fine-scale variations (<100kb) -Individual variation in hotspot intensity (sperm typing) (Jeffreys 1998, 2002, 2005) -Inter-species variation in hotspot intensity and/or location (LD) (Wall 2003, Ptak 2004, 2005, Winckler 2005) -Population differences in hotspot location and/or intensity ? (Crawford 2004) -Large-scale variations (>1Mb) -Individual variation in number of recombination events per meiosis (Yu 1996, Broman 1998, Kong 2002, 2004) -Inter-species variation in genetic map length (Rogers 2000) Are large-scale recombination patterns conserved among human populations ? Recombination variations in Humans

41. Used genome-wide polymorphism data from Perlegen Sciences (Hinds 2005) re-sequencing followed by genotyping >1,500,000 SNPs (average: one SNP per 1.8kb) 23 individuals from 3 populations - European-Americans (Utah residents with ancestry from northern and western Europe), - African-Americans, - and Han Chinese (from the Los Angeles area). Materiel & Methods

42. For each population, split genome in 1Mb-window for each window, if more than 100 SNPs, calculated ρ cl =4N e c using Hudson 2001 I considered only window for which I obtained ρ cl in each population:  2,613 non-overlapping windows of 1Mb covering the 22 autosomal chromosomes Materiel & Methods

44. Is ρ more strongly correlated with genetic map estimate in Eur.-Am.? Kong (2002) estimated recombination rates using 146 Icelandic families. If the recombination rate patterns differ between human populations, the DeCode estimates should predict better the population recombination rates in populations of European ancestry than they do in populations of non-European ancestry. N.B. the slope of the correlation is proportional to N e and will therefore differ for different populations but the strength of the correlation (i.e. Pearson’s r) will allow detecting differences in the recombination patterns. Performed a linear regression analysis between: the recombination rates estimated from polymorphism data (ρ) and those obtained in the DeCode map (c map ) for 2,609 non-overlapping 1Mb windows.

45. Pearson’s r r2r2 99% CI European-Americans0.5087[0.4733-0.5431] Han Chinese0.4986[0.4624-0.5329] African-Americans0.4964[0.4610-0.5314] Is ρ more strongly correlated with genetic map estimate in Eur.-Am.?

46. Is the ratio ρ/ρ constant along the chromosomes? Population recombination rate: ρ=4N e c If c is identical in different populations, ρ (pop1) /ρ (pop2) = N e(pop1) /N e(pop2) = constant Thus, if the patterns of recombination rates are conserved, the ratios of the population recombination rate estimates should be constant along the chromosomes. (this is true if the variations of N e along the chromosomes are negligible at the scale considered)

47. Is the ratio ρ/ρ constant along the chromosomes?

48. Looking at the most extreme ρ/ρ ratios… Looked at the 2.5% highest-2.5% lowest ratios of each comparison - no significant deviation from a uniform distribution across chromosomes - distribution in centromeric, telomeric and rest of the genome is marginally significant.

49. Looking at the most extreme ρ/ρ ratios… Is there any biological outliers? Selected regions with 2+ extreme windows within 3Mb (and in the same direction) The region with the most extreme ratios corresponds to a well-described polymorphic inversion on chromosome 8p (Giglio 2001) African-American/HanChinese European-American/HanChinese African-American/European-American 6 6 7 1 2

50. Polymorphic inversion on chromosome 8p

51. Conclusion: Overall conservation of large-scale recombination rate patterns among human populations. - Genetic maps established in European populations can be directly used for linkage analyses in other populations. - Local recombination rate differences can be used to investigate population differences: chromosomal rearrangements or selection. - There is a discrepancy between the variability of hotspots and the conservation of the recombination rates on a larger scale. It is possible that the position and intensity of recombination hotspots can vary among humans but that the overall frequency of hotspots over a larger scale has to be maintained.