Questions about MR-JTI

[novnex-cmd]

Hi @egamazon @zdangm @gamazonlab

Thank you for sharing such a wonderful job. I would like to run MR-JTI analysis but I don't know how to explain the output file. The first line "expression ... " represents the effect of the expression on the trait. So does the second line “ldsc..." represents the heterogeneity estimates?

By the way, I wonder to know whether it is appropriate to include only SNPs with fdr<0.05 in GTEx for MR-JTI analysis. look forward for your help.

Many thanks.

Best regards,

Minglong Cai

[zdangm]

Hi Minglong,
Thank you for using MR-JTI!

The first line "expression ... " represents the effect of the expression on the trait.
Correct.

So does the second line “ldsc..." represents the heterogeneity estimates?
The second line represents the effect of the LD-score which is considered as a covariate here. The following rows (one raw one SNP with potential pleiotropy effect) show the heterogeneity estimates. i.e., any SNP showing a significant 'extra effect' would result in heterogeneity.

whether it is appropriate to include only SNPs with fdr<0.05 in GTEx for MR-JTI analysis
Sure. However, it needs at least 20 independent IVs (SNPs) to have a robust estimate. Very few genes could have > 20 independent IVs. So, we suggest using all the cis SNPs.

Please let me know if you have any questions.
Best,
Dan

[novnex-cmd]

Thank you for your help.

I have one more question. you recommend providing LD Score when using MR-JTI. Should we include all GTEx SNP or pruned SNP to calculate LD score? Look forward for your help.

Many thanks.

Minglong Cai

[zdangm]

Hi Minglong, All the SNPs should be included for LD score calculation. Please let me know if there is anything else I can help with. Dan

…

On Thu, Aug 19, 2021 at 8:29 AM novnex-cmd ***@***.***> wrote: Thank you for your help. I have one more question. you recommend providing LD Score when using MR-JTI. Should we include all GTEx SNP or pruned SNP to calculate LD score? Look forward for your help. Many thanks. Minglong Cai — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#7 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AH5HAMKV67LNNIKIGTSABQLT5UBKZANCNFSM5BXDANXQ> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email> .