search

bulik / ldsc

LD Score Regression (LDSC)
GNU General Public License v3.0
601 stars 331 forks source link

p values in version Version 1.0.1 are different from v 2.0.0 #430

Open zara03 opened 1 month ago

zara03 commented 1 month ago

Hi My ex lab member did analysis on same datasets but it was version 1.0.1. I am repeating the analysis with v 2.0.0 , rg values come similar but p values which were significant with version 1.0.1 are non significant now. Can someone tell me what I might be missing?like why p values are non significant or is something which i missed in code.

aksarkar commented 3 weeks ago

@zara03 The p-values are probably different because the jacknife standard errors are different.

zara03 commented 3 weeks ago

great. Can you please tell le one more thing when should I use intercept flag ?

Get Outlook for iOShttps://aka.ms/o0ukef

From: Abhishek Sarkar @.> Sent: Sunday, May 5, 2024 4:12:47 PM To: bulik/ldsc @.> Cc: Syeda, Zehra @.>; Mention @.> Subject: Re: [bulik/ldsc] p values in version Version 1.0.1 are different from v 2.0.0 (Issue #430)

@zara03https://github.com/zara03 The p-values are probably different because the jacknife standard errors are different.

— Reply to this email directly, view it on GitHubhttps://github.com/bulik/ldsc/issues/430#issuecomment-2094934236, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AW7NJLEJ6YDMGTQWP63ANALZA2HD7AVCNFSM6AAAAABGTV3Y32VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAOJUHEZTIMRTGY. You are receiving this because you were mentioned.Message ID: @.***>

aksarkar commented 1 week ago

@zara03 The only reason to use the intercept flag is if you believe you know what the intercept is supposed to equal. This is almost never true.