search

n-mounier / MRlap

R package to perform two-sample Mendelian Randomisation (MR) analyses using (potentially) overlapping samples
GNU General Public License v2.0
51 stars 11 forks source link

No IV left after thresholding #21

Open xz-xxz opened 3 months ago

xz-xxz commented 3 months ago

Hi, I'm trying to use MRlap to support my findings. When using function 'TwosampleMR', more than 100 snps remained, but error occurred when running IVW MR using 'MRlap' as "no IV left after thresholding (5e-08). Please advise, thanks. 屏幕截图 2024-08-02 201919

yangyizhixiaomaomi commented 2 months ago

Have you found a solution to ldsc that Error in h(simpleError(msg, call)) : 在为't'函数选择方法时评估'x'参数出了错: 非整合参数

xz-xxz commented 2 months ago

This problem could be solved by using the full GWAS statistics. Have you try? Additionally, I still have this problem, as no IV left after thresholding.

yangyizhixiaomaomi commented 2 months ago

yes when I use the whole gwas ，it successd.you can try .for another problem ,l haven't . maybe your p_value too strict? l don't know.

---Original--- From: @.> Date: Sun, Aug 25, 2024 13:42 PM To: @.>; Cc: @.**@.>; Subject: Re: [n-mounier/MRlap] No IV left after thresholding (Issue #21)

This problem could be solved by using the full GWAS statistics. Have you try? Additionally, I still have this problem, as no IV left after thresholding.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

xz-xxz commented 2 months ago

@lcpilling Hi. I still have this problem. "TwosampleMR" identified more than 100 SNPs using threshold of 5e-08 and clumping of 10,000kb (r2=0.001), but "MR lap" showed no IV left when using the same parameters. Additionally, each run costs about 30 mins. Could you please check this, I don't know how to solve this error. 屏幕截图 2024-08-25 175827

lcpilling commented 2 months ago

Hi @xz-xxz - sorry to ask but are you certain the object LST contains significant (p<5e-08) variants? Could you check with table( LST$P < 5e-08 )

xz-xxz commented 2 months ago

@lcpilling Thanks for your reply. Yes, definitely, a total of 7462 variants remained after thresholding of 5e-08. 屏幕截图 2024-08-28 162602

xz-xxz commented 2 months ago

屏幕截图 2024-08-28 164017

lcpilling commented 2 months ago

OK. MRlap doesn't actually use the p-value column but instead uses the BETA and SE cols to compute the z-statistic. Perhaps there is a problem here with your input data. Could you try: table( abs(LST$BETA/LST$SE) > 5.45131) to tell you how many variants have a GWAS significant z-statistic?

xz-xxz commented 2 months ago

A total of 7438 variants have significant z-statistic. 屏幕截图 2024-08-28 171732

lcpilling commented 2 months ago

OK.... I do not know why MRlap finds no IVs - that is unusual!

xz-xxz commented 2 months ago

Well… thank you anyway！

徐振 @.***

签名由 网易灵犀办公 定制

Original: From: Luke C. @.> Date: 2024-08-28 19:18:43 To: @.> Cc: @.>, @.> Subject: Re: [n-mounier/MRlap] No IV left after thresholding (Issue #21) OK.... I do not know why MRlap finds no IVs - that is unusual! — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.***>