Handling case-control sum stats

[davidaknowles]

munge_sumstats from LDSC handles (log) odds ratios in place of beta. Is it intentional to not handle this in PolyFun?

Thanks.

[omerwe]

Hi, Good point --- I've modified the code to accept column name OR (in practice the code only uses this to determine the sign of the effect --- the Z-score is determined from the p-value).

…

On Sun, Feb 16, 2020 at 11:45 AM David A Knowles ***@***.***> wrote: munge_sumstats from LDSC handles (log) odds ratios in place of beta. Is it intentional to not handle this in PolyFun? Thanks. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#21?email_source=notifications&email_token=ACNCB47Z6QLV6HZPAQQZB6DRDFUR5A5CNFSM4KWEZIJKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4IN3OW2Q>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACNCB43ULYCIWXPKGIFZOR3RDFUR5ANCNFSM4KWEZIJA> .

[davidaknowles]

Thanks. Just to confirm, this will correctly handled OR as opposed to log odds?

[omerwe]

Thanks for catching that. I previously assumed log-odds, but now the code should convert regular odds-ratio to log-odds (we only need this to determine the sign of the effect).

…

On Mon, Feb 17, 2020 at 3:46 PM David A Knowles ***@***.***> wrote: Thanks. Just to confirm, this will correctly handled OR as opposed to log odds? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#21?email_source=notifications&email_token=ACNCB4ZE3MJCDKQKEGED3ZDRDLZSBA5CNFSM4KWEZIJKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEL7TW4A#issuecomment-587152240>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACNCB47OPRBXJLWETIFSDALRDLZSBANCNFSM4KWEZIJA> .

[davidaknowles]

Great, thanks.