search

MRCIEU / TwoSampleMR

R package for performing 2-sample MR using MR-Base database
https://mrcieu.github.io/TwoSampleMR
Other
433 stars 176 forks source link

Inconsistent results using MRC-IEU GWAS (extracting outcomes) #55

Open jonnyrose opened 5 years ago

jonnyrose commented 5 years ago

I ran two sample MR on a series of alcohol related outcomes on Dec 26. I saved all the results from extracting outcomes to mr results (including mr heterogenity, mr pleiotropy, mr singlesnp, mr loo).

I went back today to check my work.

On Dec 26, I used an exposure data set of 70 snps. Extracting outcomes from b:1707, 5779, 5174, etc, I got 68 snps plus 2 proxies (total 70 snps).

Today, exact same code, exact same exposure data set of 70 snps, rerun:

Extracting data for 70 SNP(s) from 1 GWAS(s) Token cache file: mrbase.oauth Finding proxies for 21 SNPs in outcome UKB-b:5359 Extracting data for 21 SNP(s) from 1 GWAS(s)

8 proxies were apparently not found. Only 62 came back (with 13 proxies).

This is the same for all of the alcohol related MRC-IEU GWAS.

Have you modified the GWAS? or the code? Thanks in advance for any help you can give me here.

explodecomputer commented 5 years ago

Hi @jonnyrose thanks for letting us know about this. I'm not sure why this might have happened, could you please try rolling back the TwoSampleMR version

devtools::install_github("MRCIEU/TwoSampleMR@0.4.17")

or 0.4.18 and seeing if you can recover the old behaviour? If not then we'll look at other possible causes.

Thanks

jonnyrose commented 5 years ago

I am using 4.16, not sure why except at some point there was some message suggesting rolling back to that version. When I reran my code this morning, I got the first set of results (70 snps, 68 found, 2 proxies). (When I said it was the same for all alcohol related phenotypes, I meant that I got the same results trying to extract instruments (using the same exposure data) for all phenotypes.) Thanks anyway.

jonnyrose commented 5 years ago

UPDATE - When I reran the code this evening, I get the second set of results (21 proxies) instead of the first set. Again, I am using version 4.16. It happens for more than one outcome. Actually, I tried to do the same on MR Base to check but the app kept kicking me out when I hit run MR. Any help here would be appreciated

explodecomputer commented 5 years ago

Thanks @jonnyrose. Is there a chance you could send us the list of SNPs, this will help us try to track down what is happening. If it’s sensitive info you can send to my email directly if you prefer

From: jonnyrose notifications@github.com Reply-To: MRCIEU/TwoSampleMR reply@reply.github.com Date: Monday, 21 January 2019 at 21:16 To: MRCIEU/TwoSampleMR TwoSampleMR@noreply.github.com Cc: gibran hemani explodecomputer@gmail.com, Comment comment@noreply.github.com Subject: Re: [MRCIEU/TwoSampleMR] Inconsistent results using MRC-IEU GWAS (extracting outcomes) (#55)

I am using 4.16, not sure why except at some point there was some message suggesting rolling back to that version. When I reran my code this morning, I got the first set of results (70 snps, 68 found, 2 proxies). (When I said it was the same for all alcohol related phenotypes, I meant that I got the same results trying to extract instruments (using the same exposure data) for all phenotypes.) Thanks anyway.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

ruchika1227 commented 7 months ago

Hi, I am getting an error 400 to use MR Base -- is anyone else receiving this error?