Open jonnyrose opened 5 years ago
Hi Jonnyrose
I recommend that the function is only used where the summary data has been derived from a logistic regression model of a binary phenotype.
bw
Philip
From: jonnyrose notifications@github.com Sent: 24 March 2019 17:28:01 To: MRCIEU/TwoSampleMR Cc: Subscribed Subject: [MRCIEU/TwoSampleMR] New generate odds ratio function (#65)
Just to be sure, I understand this function is useful when outcome is a binary phenotype, where GWAS association method was logistics regression. What if the outcome is the ordered categorical type (e.g. frequency outcomes: never = 0, 1-2 times = 1, 3-6 times = 2, etc), but GWAS association method used was linear regression; the beta in the summary statistics would not be log odds of the outcome. Would it be valid to use generate_odds_ratio function in any case other than a binary phenotype from a GWAS estimated using logistic regression?
Thanks in advance for any help you can give me.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/MRCIEU/TwoSampleMR/issues/65, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AKQwPk_Y1aN_YR_TuB7qHqnJ_u-rHr9Kks5vZ7WhgaJpZM4cFoWq.
Just to be sure, I understand this function is useful when outcome is a binary phenotype, where GWAS association method was logistics regression. What if the outcome is the ordered categorical type (e.g. frequency outcomes: never = 0, 1-2 times = 1, 3-6 times = 2, etc), but GWAS association method used was linear regression; the beta in the summary statistics would not be log odds of the outcome. Would it be valid to use generate_odds_ratio function in any case other than a binary phenotype from a GWAS estimated using logistic regression?
Thanks in advance for any help you can give me.