diff options
author | Alexander_Kabui | 2024-01-02 13:21:07 +0300 |
---|---|---|
committer | Alexander_Kabui | 2024-01-02 13:21:07 +0300 |
commit | 70c4201b332e0e2c0d958428086512f291469b87 (patch) | |
tree | aea4fac8782c110fc233c589c3f0f7bd34bada6c /gn2/maintenance/get_group_samplelists.py | |
parent | 5092eb42f062b1695c4e39619f0bd74a876cfac2 (diff) | |
parent | 965ce5114d585624d5edb082c710b83d83a3be40 (diff) | |
download | genenetwork2-70c4201b332e0e2c0d958428086512f291469b87.tar.gz |
merge changes
Diffstat (limited to 'gn2/maintenance/get_group_samplelists.py')
-rw-r--r-- | gn2/maintenance/get_group_samplelists.py | 47 |
1 files changed, 47 insertions, 0 deletions
diff --git a/gn2/maintenance/get_group_samplelists.py b/gn2/maintenance/get_group_samplelists.py new file mode 100644 index 00000000..6af637ea --- /dev/null +++ b/gn2/maintenance/get_group_samplelists.py @@ -0,0 +1,47 @@ +import os +import glob +import gzip + +from gn2.base import webqtlConfig + + +def get_samplelist(file_type, geno_file): + if file_type == "geno": + return get_samplelist_from_geno(geno_file) + elif file_type == "plink": + return get_samplelist_from_plink(geno_file) + + +def get_samplelist_from_geno(genofilename): + if os.path.isfile(genofilename + '.gz'): + genofilename += '.gz' + genofile = gzip.open(genofilename) + else: + genofile = open(genofilename) + + for line in genofile: + line = line.strip() + if not line: + continue + if line.startswith(("#", "@")): + continue + break + + headers = line.split("\t") + + if headers[3] == "Mb": + samplelist = headers[4:] + else: + samplelist = headers[3:] + return samplelist + + +def get_samplelist_from_plink(genofilename): + genofile = open(genofilename) + + samplelist = [] + for line in genofile: + line = line.split(" ") + samplelist.append(line[1]) + + return samplelist |