diff options
| author | Hao Chen | 2020-04-22 08:24:19 -0500 |
|---|---|---|
| committer | Hao Chen | 2020-04-22 08:24:19 -0500 |
| commit | 5acb78ce96b7f47f0827838a5cc58d0527953f74 (patch) | |
| tree | 8557ed9f7c1b26dd40667ecc8189143c48ad8123 /utility/process_gwas.py | |
| parent | 2f58b4d9989cd319cca86b24edcc6eaa45a49c7b (diff) | |
| parent | 4999cb684b2db9ddc16dbbbbf600df0ff947ae07 (diff) | |
| download | genecup-5acb78ce96b7f47f0827838a5cc58d0527953f74.tar.gz | |
Merge branch 'master' of https://github.com/chen42/atpub
Diffstat (limited to 'utility/process_gwas.py')
| -rw-r--r-- | utility/process_gwas.py | 48 |
1 files changed, 48 insertions, 0 deletions
diff --git a/utility/process_gwas.py b/utility/process_gwas.py new file mode 100644 index 0000000..eba59c0 --- /dev/null +++ b/utility/process_gwas.py @@ -0,0 +1,48 @@ +import re + +with open("./addiction_gwas.tsv", "r") as f: + for line in f: + try: + (pmid, trait0, gene0, gene1, snp, pval, trait1)=line.strip().split("\t") + except: + next + key1="unassigned" + key2="unassigned" + trait=trait0+"; "+trait1 + genes=gene0+";"+gene1 + if re.search('cocaine', trait, flags=re.I): + key1="addiction" + key2="cocaine" + elif re.search('smoking|congestive|nicotine', trait, flags=re.I): + key1="addiction" + key2="nicotine" + elif re.search('opioid|morphin|heroin|methadone', trait, flags=re.I): + key1="addiction" + key2="opioid" + elif re.search('amphetam', trait, flags=re.I): + key1="addiction" + key2="amphetamine" + elif re.search('canabis', trait, flags=re.I): + key1="addiction" + key2="canabis" + elif re.search('food', trait, flags=re.I): + key1="addiction" + key2="food" + elif re.search('alcohol', trait, flags=re.I): + key1="addiction" + key2="alcohol" + elif re.search('addiction|abuse', trait, flags=re.I): + key1="addiction" + key2="addiction" + else: + key1="behavior" + key2="psychiatric" + genes=genes.replace(" - ", ";") + genes=genes.replace(",", ";") + printed=dict() + for gene in genes.split(";"): + gene=gene.replace(" ","") + if gene !="NR" and gene not in printed: + text="SNP:<b>"+snp+"</b>, P value: <b>"+pval+"</b>, Disease/trait:<b> "+trait0+"</b>, Mapped trait:<b> "+trait1+"</b>" + print (gene+"\t"+"GWAS"+"\t"+key2+"_GWAS\t"+pmid+"\t"+text) + printed[gene]=1 |