1 / 29

Gene-Environment, Gene-Gene Interaction Quanto (power calculation) hydrac/gxe/

Lab 3. Gene-Environment, Gene-Gene Interaction Quanto (power calculation) http://hydra.usc.edu/gxe/. Yu-Chun Jean Yen yyen@hsph.harvard.edu Bldg.2 Rm. 200. Sample Dataset Yahoo06.dat: a comma delimited file Case-control with 1680 subjects, 16 SNPs. data yahoo; set yahoo;

aquarius
Download Presentation

Gene-Environment, Gene-Gene Interaction Quanto (power calculation) hydrac/gxe/

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Lab 3 • Gene-Environment, Gene-Gene Interaction • Quanto (power calculation)http://hydra.usc.edu/gxe/ Yu-Chun Jean Yenyyen@hsph.harvard.eduBldg.2 Rm. 200

  2. Sample Dataset Yahoo06.dat: a comma delimited file Case-control with 1680 subjects, 16 SNPs

  3. data yahoo; set yahoo; /* define genotype scores for snps 5 and 10 */ z5=(a9 eq 3)+(a10 eq 3); if a9 eq . or a10 eq .then z5=.; z10=(a19 eq 4)+(a20 eq 4); if a19 eq . or a20 eq .then z10=.; run;

  4. /* "interaction tests" assuming linear trend in log OR by SNP5 */ /* Get null -2 log L for "joint test" */ proclogisticdata=yahoo; where z5 ne . and x ne .; *complete case subset; model d(event='2')=x; /* Get null -2 log L for "test of statistical interaction" */ proclogisticdata=yahoo; where z5 ne . and x ne .; model d(event='2')=x z5; proclogisticdata=yahoo; where z5 ne . and x ne .; model d(event='2')=x z5 z5*x; run;

  5. Test for SNP5 (adjusted for X): LRT=9.2 on 1 d.f. p=.002(model 2 vs. modle 1) Test for "statistical interaction" LRT=0.9 on 1 d.f. p=.34(model 3 vs. model 2) Joint test for G and G-E effects LRT=10.1 on 2 d.f p=.006(model 3 vs. model 1)

  6. /* g-g "interaction" test assuming linear trend in log OR by SNP5 and SNP10 */ proclogisticdata=yahoo; where z5 ne . and z10 ne .; model d(event='2')=z5 z10 z5*z10; run; /* g-g "interaction" test not assuming linear trend in OR by SNP5 and SNP10 */ proclogisticdata=yahoo; where z5 ne . and z10 ne .; class z5 z10 (ref='0'param=reference); model d(event='2')=z5 z10 z5*z10; run;

  7. /* g-g "interaction" test not assuming linear trend in OR by SNP5 and SNP10 */ data yahoo; set yahoo; Z5_Z10=(put(z5, 1.))||(put(z10, 1.)); if z5=. or z10=.then Z5_Z10=' '; run; procfreqdata=yahoo; tables z5_z10*z5*z10 / list; run; proclogisticdata=yahoo; where z5 ne . and z10 ne .; class z5_z10 (ref='00'param=reference); model d(event='2')=z5_z10; run;

  8. /* fit (constrained) additive interaction between dichotomous g and x */ data yahoo; set yahoo; if z5 eq 0then gcat=0; if z5 gt 0then gcat=1; if x le .5 and x ne .then xcat=0; if x gt .5then xcat=1; procfreqdata=yahoo; tables gcat*xcat / list; run; procnlmixed data=yahoo; where gcat ne.and xcat ne.; if (gcat eq0) and (xcat eq0) then eta=a; if (gcat eq0) and (xcat eq1) then eta=a+b2; if (gcat eq1) and (xcat eq0) then eta=a+b1; if (gcat eq1) and (xcat eq1) then eta=a+log(exp(b1)+exp(b2)-1); ll = (d=2)*eta - log(1+exp(eta)); model d ~ general(ll); parms a b1 b2=0; run;

  9. Beta_2 is log OR (exposed non carriers) Beta_1 is log OR (carrier unexposed) OR (exposed carrier) is exp(beta_1)+exp(beta_2)-1

  10. /* unconstrained model (to test for departure from additive--risk scale--interaction */ proclogisticdata=yahoo; model d(event='2')=gcat xcat gcat*xcat; run; proclogisticdata=yahoo; class G_X (ref='00'param=reference); model d(event='2')=G_X; run; procnlmixed data=yahoo; where gcat ne.and xcat ne.; if (gcat eq0) and (xcat eq0) then eta=a; if (gcat eq0) and (xcat eq1) then eta=a+b2; if (gcat eq1) and (xcat eq0) then eta=a+b1; if (gcat eq1) and (xcat eq1) then eta=a+b3; ll = (d=2)*eta - log(1+exp(eta)); model d ~ general(ll); parms a b1 b2 b3=0; run;

  11. Constrained -2 log L = 2211.7 Unconstrained -2 log L = 2210.8 LRT is 0.9 on 1 d.f. p=0.34 Beta_2 is log OR (exposed non carriers) Beta_1 is log OR (carrier unexposed) Beta_3 is log OR (exposed carrier)

  12. Power calculations: Quanto http://hydra.usc.edu/gxe/

  13. G, G-E, G-G, joint G G-E Set MAF Set Prev(E) Set Penetrance Model Set alpha, target power (N)

More Related