use https://raw.github.com/remlapmot/mrrobust/master/dodata, clearExample demonstrating MVMR commands
Read in the Do et al. (2013) example dataset.
Select observations (p-value with LDL-C < 10-8)
gen byte sel1 = (ldlcp2 < 1e-8)MV-IVW
Fit the multivariable inverse-variance weighted (MV-IVW a.k.a. multivariable Mendelian randomization, MVMR) estimator with phenotypes LDL-c and HDL-c (Burgess, Dudbridge, and Thompson 2015).
mrmvivw chdbeta ldlcbeta hdlcbeta [aw=1/(chdse^2)] if sel1==1 Number of genotypes = 73
Number of phenotypes = 2
Standard errors: Random effect
Residual standard error = 1.514
------------------------------------------------------------------------------
| Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
chdbeta |
ldlcbeta | .4670719 .0581901 8.03 0.000 .3530214 .5811224
hdlcbeta | -.2930048 .1211822 -2.42 0.016 -.5305175 -.0554921
------------------------------------------------------------------------------Additionally include a third phenotype – triglycerides.
mrmvivw chdbeta ldlcbeta hdlcbeta tgbeta [aw=1/(chdse^2)] if sel1==1 Number of genotypes = 73
Number of phenotypes = 3
Standard errors: Random effect
Residual standard error = 1.490
------------------------------------------------------------------------------
| Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
chdbeta |
ldlcbeta | .42862 .0609661 7.03 0.000 .3091286 .5481113
hdlcbeta | -.1941989 .1308289 -1.48 0.138 -.4506189 .0622211
tgbeta | .2260456 .1232828 1.83 0.067 -.0155842 .4676755
------------------------------------------------------------------------------Report the QA statistic for instrument validity and the conditional F-statistics for instrument strength for each phenotype (Sanderson et al. 2019; Sanderson, Spiller, and Bowden 2021).
mrmvivw chdbeta ldlcbeta hdlcbeta tgbeta [aw=1/(chdse^2)] if sel1==1, gxse(ldlcse hdlcse tgse) Number of genotypes = 73
Number of phenotypes = 3
Standard errors: Random effect
Residual standard error = 1.490
------------------------------------------------------------------------------
| Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
chdbeta |
ldlcbeta | .42862 .0609661 7.03 0.000 .3091286 .5481113
hdlcbeta | -.1941989 .1308289 -1.48 0.138 -.4506189 .0622211
tgbeta | .2260456 .1232828 1.83 0.067 -.0155842 .4676755
------------------------------------------------------------------------------
Q_A statistic for instrument validity; chi2(70) = 152.88 (p = 0.0000)
Conditional F-statistics for instrument strength:
F_x1 = 130.31 (ldlcbeta)
F_x2 = 36.29 (hdlcbeta)
F_x3 = 40.44 (tgbeta)MVMR-Egger
Fit MVMR-Egger regression (Rees, Wood, and Burgess 2017), by default orienting the model to the first phenotype in the main varlist.
mrmvegger chdbeta ldlcbeta hdlcbeta tgbeta [aw=1/(chdse^2)] if sel1==1 MVMR-Egger model oriented wrt: ldlcbeta
Number of genotypes = 73
Number of phenotypes = 3
Residual standard error = 1.469
------------------------------------------------------------------------------
| Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
chdbeta |
ldlcbeta | .5672993 .1002611 5.66 0.000 .370791 .7638075
hdlcbeta | -.1364113 .1332727 -1.02 0.306 -.3976209 .1247983
tgbeta | .2739803 .1246927 2.20 0.028 .0295871 .5183735
_cons | -.0093655 .0054187 -1.73 0.084 -.019986 .001255
------------------------------------------------------------------------------We can also orient the model with respect to HDL-C instead of LDL-C.
mrmvegger chdbeta ldlcbeta hdlcbeta tgbeta [aw=1/(chdse^2)] if sel1==1, orient(2) MVMR-Egger model oriented wrt: hdlcbeta
Number of genotypes = 73
Number of phenotypes = 3
Residual standard error = 1.501
------------------------------------------------------------------------------
| Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
chdbeta |
ldlcbeta | .4286398 .0614056 6.98 0.000 .308287 .5489926
hdlcbeta | -.1989637 .1541909 -1.29 0.197 -.5011723 .1032449
tgbeta | .2256794 .1243221 1.82 0.069 -.0179875 .4693463
_cons | .0002155 .0036218 0.06 0.953 -.006883 .0073141
------------------------------------------------------------------------------Or we can orient the model with respect to triglycerides instead of LDL-C.
mrmvegger chdbeta ldlcbeta hdlcbeta tgbeta [aw=1/(chdse^2)] if sel1==1, orient(3) MVMR-Egger model oriented wrt: tgbeta
Number of genotypes = 73
Number of phenotypes = 3
Residual standard error = 1.499
------------------------------------------------------------------------------
| Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
chdbeta |
ldlcbeta | .4203073 .0660026 6.37 0.000 .2909447 .54967
hdlcbeta | -.1903089 .1321536 -1.44 0.150 -.4493252 .0687075
tgbeta | .2065651 .1365427 1.51 0.130 -.0610537 .474184
_cons | .0013499 .003951 0.34 0.733 -.0063939 .0090936
------------------------------------------------------------------------------References
Burgess, S, F Dudbridge, and SG Thompson. 2015. “Multivariable Mendelian randomization: the use of pleiotropic genetic variants to estimate causal effects.” American Journal of Epidemiology 181 (4): 251–60. https://doi.org/10.1093/aje/kwu283.
Do, Ron, Cristen J Willer, Ellen M Schmidt, Sebanti Sengupta, Chi Gao, Gina M Peloso, Stefan Gustafsson, et al. 2013. “Common Variants Associated with Plasma Triglycerides and Risk for Coronary Artery Disease.” Nature Genetics 45 (11): 1345–52. https://doi.org/10.1038/ng.2795.
Rees, J, A Wood, and S Burgess. 2017. “Extending the MR-Egger method for multivariable Mendelian randomization to correct for both measured and unmeasured pleiotropy.” Statistics in Medicine 36 (29): 4705–18. https://doi.org/10.1002/sim.7492.
Sanderson, E, G Davey Smith, F Windmeijer, and J Bowden. 2019. “An examination of multivariable Mendelian randomization in the single-sample and two-sample summary data settings.” International Journal of Epidemiology 48 (3): 713–27. https://doi.org/10.1093/ije/dyy262.
Sanderson, E, W Spiller, and J Bowden. 2021. “Testing and correcting for weak and pleiotropic instruments in two-sample multivariable Mendelian randomization.” Statistics in Medicine 40 (25): 5434–52. https://doi.org/10.1002/sim.9133.