huwenboshi/hess

Dealing with loci with no SNP

Opened this issue · 4 comments

ggittu commented

Hi Huwenbo-

@shihuwenbo @huwenboshi I am trying to work this out for the step2, but I am getting error ,

[INFO] Loaded results for 1703 loci from step 1
[INFO] Using 1066727 SNPs with average sample size 8965.0
[INFO] Re-inflate the summary statistics with lambda_gc: 1
[ERROR] Rank of A less than the number of loci. There might be loci with no SNP.

From what is given in the FAQS, I made sure the sumstats file dont have any NA in the SNP column. What can I do to make this run ? Which SNP's from where should be removed? Is the --drop-missing-loci function implemented yet ?

Thanks Huwenbo

meghatron21 commented

Hi. I wanted to follow up on this also. I am getting an error of Rank of A less than number of loci. Should I just use a different partiion function?

Rick-Chen-PKU commented

Getting the same error! Any advice to handle it?

Rick-Chen-PKU commented

I may have found the problem, there is little SNPs in a region on chr1:178944309-178954470 (see below). If use sumstat.gz data from LDSC to run the code, it may pop out an error!
The error will cause subsequent errors [ERROR] Rank of A less than the number of loci. There might be loci with no SNP.
70712433e36db8f1ace3d3a70caa4c6
d49b487412af0bb17e9c75fb6ab52e5

huwenboshi commented

Hi, in this case, you will need to remove the region in the bed file, and rerun the analysis.