search

omerwe / polyfun

PolyFun (POLYgenic FUNctionally-informed fine-mapping)
MIT License
96 stars 22 forks source link

'nan' in LD matrix (.bcor)/ #163

Closed chunt28 closed 1 year ago

chunt28 commented 1 year ago

I am running susie with the finemapper.py script using a pre-calculated LD matrix for my region of interest. I get the following error message.

Traceback (most recent call last): File "/home/c/ch553/polyfun/finemapper.py", line 1275, in df_finemap = finemap_obj.finemap(locus_start=args.start, locus_end=args.end, num_causal_snps=args.max_num_causal, File "/home/c/ch553/polyfun/finemapper.py", line 745, in finemap ld_arr, df_ld_snps = read_ld_from_file(ld_file) File "/home/c/ch553/polyfun/finemapper.py", line 120, in read_ld_from_file ld_arr, df_ld_snps = load_ld_bcor(ld_file[:-5]) File "/home/c/ch553/polyfun/finemapper.py", line 103, in load_ld_bcor assert np.all(~np.isnan(ld_arr)) AssertionError

It seems that the issue arises because some SNP correlations are labelled as 'nan' in my bcor files. If this is the case, what is the best way to deal with this issue.

Thank you

omerwe commented 1 year ago

Hi @chunt28 , as you say, it sounds like something is wrong in the bcor file. Without knowing more I can't help. If you can send a full end-to-end reproducible example (including how you ran FINEMAP), I might know more. Sorry I can't help more!

chunt28 commented 1 year ago

I discovered that the missingness threshold set by LDstore was resulting in some SNP correlations being labelled as 'nan' in the matrix, but I have resolved the issue.

Thanks