Closed jerome-f closed 3 years ago
Hi,
I specifically made finemapper.py append a zero because that's what LDStore does, and I wanted to make them consistent with each other. Can you please describe how you created a bcor file that doesn't have a preceding zero?
Hi Omer,
I created the .bcor files independent of the finemapper.py. When I created the .bcor using ldstore the .z file listing all the snps had chromosome 1-9 as single characters. I am not sure if specifying chromosome as 01 instead of 1 is a convention (I am fairly new to this)
Best Jerome
Hi Jerome,
This is actually not a convention --- it's just a property of FINEMAP+LDStore. FINEMAP doesn't work if chromosome numbers are specified without a trailing zero. Can you please send me the exact set of commands you used to generate a .bcor file + apply FINEMAP to run into this error? If would be best if you can use the plink files in the PolyFun example files so that I can reproduce this exactly.
Thanks!
Omer
Hi Omer,
Sorry for the late replies, I am not able to share my genotype files due to restrictions. the command I used is
ldstore_v2.0_x86_64 --in-files chr2_166000001_169000001.dat --write-bcor --read-bdose --n-threads 4 --memory 24
The .dat file has all the following information
z;bgen;bgi;sample;bcor;bdose;ld;n_samples
chr2_166000001_169000001.z;xxx.bgen;xxx.bgen.bgi;xxx.sample;chr2_166000001_169000001.bcor;xxx.bdose;xxx.ld;332423
the chr2_166000001_169000001.z has the following information
rs26 2 168743403 T A
rs29 2 168743545 C A
rs30 2 168743603 C A
rs31 2 168743611 G T
rs32 2 168743662 T C
here the second column is just one character, and it gets stored in the .bcor file as such. I can confirm that if I change the chr2_166000001_169000001.z as
rs26 02 168743403 T A
rs29 02 168743545 C A
rs30 02 168743603 C A
rs31 02 168743611 G T
rs32 02 168743662 T C
where the second column is 2 chars the bcor file also ends up have 2 chars. So really it is not a actual issue. I curated the .z file and chose to create it with one char.
Hi Jerome,
Thanks for the explanation, I didn't realize that FINEMAP now also supports non-padded chromosome numbers (it used not to). I modified the code to (hopefully) support both representations. Can you please git pull and see if the problem is resolved now?
Hi Omer,
Just tested the code working fine now.
Hi Omer,
finemapper.py pads a 0 in front of chromosome number 1-9, this creates a snp mismatch between the bcor file and the input passed from finemapper.py script during the execution of FINEMAP. There should likely be a note in the Wiki about it. If the .bcor file is prepared independent of finemapper.py then the chromosome position can be represented as 1 or 01.