1. Power in QTL linkage: single and multilocus analysis Shaun Purcell 1,2 & Pak Sham 1 1 SGDP, IoP, London, UK 2 Whitehead Institute, MIT, Cambridge, MA, USA

2. Overview 1) Brief power primer 2) Calculating power for QTL linkage analysis Practical 1 : Using GPC for linkage power calculations 3) The adequacy of additive single locus analysis Practical 2 : Using Mx for linkage power calculations

3. 1) Power primer Test statistic YES NO

4. P(T)P(T) T Critical value Sampling distribution if H A were true Sampling distribution if H 0 were true  

5. Rejection of H 0 Nonrejection of H 0 H 0 true H A true Type I error at rate  Type II error at rate  Significant result Nonsignificant result POWER =(1-  )

6. Impact of  effect size, N P(T)P(T) T Critical value  

7. Impact of  alpha P(T)P(T) T Critical value  

8. 2) Power for QTL linkage For chi-squared tests on large samples, power is determined by non-centrality parameter ( ) and degrees of freedom (df) = E(2lnL A - 2lnL 0 ) = E(2lnL A ) - E(2lnL 0 ) where expectations are taken at asymptotic values of maximum likelihood estimates (MLE) under an assumed true model

9. Linkage test HAH0HAH0 for i=j for i  j for i=j for i  j

10. Linkage test For sib-pairs under complete marker information Expected NCP Determinant of 2-by-2 standardised covariance matrix = 1 - r 2 Note: standardised trait See Sham et al (2000) AJHG, 66. for further details

11. 200 sibling pairs; sibling correlation 0.5. To calculate NCP if QTL explained 10% variance: 200 × 0.002791 = 0.5581 Concrete example

12. Approximation of NCP NCP per sibship is proportional to - the # of pairs in the sibship (large sibships are powerful) - the square of the additive QTL variance (decreases rapidly for QTL of v. small effect) - the sibling correlation (structure of residual variance is important)

13. 0 1 2 012 IBD at QTL IBD at M P(IBD at M | IBD at QTL)

14. Using GPC Comparison to Haseman-Elston regression linkage Amos & Elston (1989) H-E regression - 90% power (at significant level 0.05) - QTL variance 0.5 - marker & major gene completely linked (θ = 0)  320 sib pairs - if θ = 0.1  778 sib pairs

15. GPC input parameters Proportions of variance additive QTL variance dominance QTL variance residual variance (shared / nonshared) Recombination fraction ( 0 - 0.5 ) Sample size & Sibship size ( 2 - 8 ) Type I error rate Type II error rate

16. GPC output parameters Expected sibling correlations - by IBD status at the QTL - by IBD status at the marker Expected NCP per sibship Power - at different levels of alpha given sample size Sample size - for specified power at different levels of alpha given power

17. GPC http://ibgwww.colorado.edu/~pshaun/gpc/

18. From GPC Modelling additive effects only SibshipsIndividuals Pairs216 (320)432 Pairs (  = 0.1)543 (778)1086 Trios (  = 0.1)179537 Quads (  = 0.1)90360 Quints (  = 0.1)55275

19. Practical 1 Using GPC, what is the effect on power to detect linkage of : 1. QTL variance? 2. residual sibling correlation? 3. marker-QTL recombination fraction?

20. Pairs required (  =0, p=0.05, power=0.8)

22. Effect of residual correlation QTL additive effects account for 10% trait variance Sample size required for 80% power (  =0.05) No dominance  = 0.1 Aresidual correlation 0.35 Bresidual correlation 0.50 Cresidual correlation 0.65

23. Individuals required

24. Effect of incomplete linkage

26. Some factors influencing power 1. QTL variance 2. Sib correlation 3. Sibship size 4. Marker informativeness & density 5. Phenotypic selection

27. Marker informativeness: Markers should be highly polymorphic - alleles inherited from different sources are likely to be distinguishable Heterozygosity (H) Polymorphism Information Content (PIC) - measure number and frequency of alleles at a locus

28. Polymorphism Information Content IF a parent is heterozygous, their gametes will usually be informative. BUT if both parents & child are heterozygous for the same genotype, origins of child’s alleles are ambiguous IF C = the probability of this occurring,

29. Singlepoint Trait locus Marker 1 θ1θ1 Trait locus Marker 1 Marker 2 T1T1 Multipoint T2T2 T3T3 T4T4 T5T5 T6T6 T7T7 T8T8 T9T9 T 10 T 11 T 12 T 13 T 14 T 15 T 16 T 17 T 18 T 19 T 20

30. Multipoint PIC: 10 cM map

31. Multipoint PIC: 5 cM map

32. The Singlepoint Information Content of the markers: Locus 1 PIC = 0.375 Locus 2 PIC = 0.375 Locus 3 PIC = 0.375 The Multipoint Information Content of the markers: PosMPIC -10 22.9946 -9 24.9097 -8 26.9843 -7 29.2319 -6 31.6665 -5 34.304 -4 37.1609 -3 40.256 -2 43.6087 -1 47.2408 0 51.1754 1 49.6898 … meaninf 50.2027

33. Selective genotyping ASPExtreme Discordant Maximally Dissimilar Proband Selection EDAC Unselected Mahanalobis Distance

34. Sibship informativeness : sib pairs

35. Impact of selection

36. E(-2LL) Sib 1 Sib 2 Sib 3 0.00121621 1.00 1.00 0.14137692-2.00 2.00 0.00957190 2.00 1.80 2.20 0.00005954 -0.50 0.50

37. 3) Single additive locus model locus A shows an association with the trait locus B appears unrelated Locus A Locus B

38. Joint analysis locus B modifies the effects of locus A: epistasis

39. Partitioning of effects Locus A Locus B MP MP

40. 4 main effects M P M P Additive effects

41. 6 twoway interactions MP MP   Dominance

42. 6 twoway interactions M PM P   Additive-additive epistasis M PP M  

43. 4 threeway interactions MPM P  M P MP   M P  MP   Additive- dominance epistasis

44. 1 fourway interaction MMP  Dominance- dominance epistasis P

45. One locus Genotypic means AAm + a Aam + d aam - a 0 d +a-a

46. Two loci AAAaaa BB Bb bb m m m m m m m m m + a A – a A + d A + a B – a B + d B – aa + aa + dd + ad – da + da – ad

47. IBD locus 1 2 Expected Sib Correlation 01  2 A /2 +  2 S 02  2 A +  2 D +  2 S 10  2 A /2 +  2 S 11  2 A /2 +  2 A /2 +  2 AA /4 +  2 S 12  2 A /2 +  2 A +  2 D +  2 AA /2 +  2 AD /2 +  2 S 20  2 A +  2 D +  2 S 21  2 A +  2 D +  2 A /2 +  2 AA /2 +  2 DA /2 +  2 S 22  2 A +  2 D +  2 A +  2 D +  2 AA +  2 AD +  2 DA +  2 DD +  2 S 002S002S

48. Estimating power for QTL models Using Mx to calculate power i. Calculate expected covariance matrices under the full model ii. Fit model to data with value of interest fixed to null value i.True modelii. Submodel Q0SN -2LL0.000=NCP

49. Model misspecification Using the domqtl.mx script i.True ii. Full iii. Null Q A Q A 0 Q D 00 SSS NNN -2LL0.000T 1 T 2 Test: dominance onlyT 1 additive & dominanceT 2 additive onlyT 2 -T 1

50. Results Using the domqtl.mx script i.True ii. Full iii. Null Q A 0.10.2170 Q D 0.100 S 0.40.3670.475 N 0.40.4170.525 -2LL0.0001.26912.549 Test: dominance only (1df)1.269 additive & dominance(2df)12.549 additive only(1df)12.549 - 1.269 = 11.28

51. Expected variances, covariances i.True ii. Full iii. Null Var1.001.00051.0000 Cov(IBD=0)0.400.3667 0.4750 Cov(IBD=1)0.450.4753 0.4750 Cov(IBD=2)0.60 0.58390.4750

52. Potential importance of epistasis “… a gene’s effect might only be detected within a framework that accommodates epistasis…” Locus A A 1 A 1 A 1 A 2 A 2 A 2 Marginal Freq.0.250.500.25 B 1 B 1 0.250010.25 Locus B B 1 B 2 0.5000.500.25 B 2 B 2 0.251000.25 Marginal0.250.250.25

53. - DDV* A1 V* D1 V* A2 V* D2 V* AA V* AD V* DA - - ADV* A1 V* D1 V* A2 V* D2 V* AA --- - AAV* A1 V* D1 V* A2 V* D2 ---- - DV* A1 -V* A2 ----- - AV* A1 ------- H0--------H0-------- FullV A1 V D1 V A2 V D2 V AA V AD V DA V DD V S and V N estimated in all models

54. True model VC Means matrix 0 0 0 0 1 1

55. NCP for test of linkage NCP1 Full model NCP2 Additive only model

56. Apparent VC under additive-only model Means matrix 0 0 0 0 1 1

57. Summary Linkage has low power to detect QTL of small effect Using selected and/or larger sibships increases power Single locus additive analysis is usually acceptable

58. GPC: two-locus linkage Using the module, for unlinked loci A and B with Means :Frequencies: 001p A = p B = 0.5 00.50 100 Power of the full model to detect linkage? Power to detect epistasis? Power of the single additive locus model? (1000 pairs, 20% joint QTL effect, VS=VN)