Mercurial > repos > davidvanzessen > imgt_loader_igg
changeset 1:4405849aa053 draft
Uploaded
| author | davidvanzessen |
|---|---|
| date | Thu, 06 Mar 2014 09:34:10 -0500 |
| parents | f8ba6ab5514b |
| children | 477cab9c32d1 |
| files | imgtconvert.py |
| diffstat | 1 files changed, 13 insertions(+), 3 deletions(-) [+] |
line wrap: on
line diff
--- a/imgtconvert.py Thu Jan 23 08:09:30 2014 -0500 +++ b/imgtconvert.py Thu Mar 06 09:34:10 2014 -0500 @@ -1,4 +1,5 @@ import pandas as pd +pd.options.mode.chained_assignment = None # default='warn' import re import argparse import os @@ -108,9 +109,18 @@ outFrame.columns = [u'ID', u'VDJ Frame', u'Top V Gene', u'Top D Gene', u'Top J Gene', u'CDR1 Seq', u'CDR1 Length', u'CDR2 Seq', u'CDR2 Length', u'CDR3 Seq', u'CDR3 Length', u'CDR3 Seq DNA', u'CDR3 Length DNA', u'Strand', u'CDR3 Found How', u'Functionality', 'V-REGION identity %', 'V-REGION identity nt', 'D-REGION reading frame', 'AA JUNCTION', 'Functionality comment', 'Sequence', 'FR1-IMGT', 'FR2-IMGT', 'FR3-IMGT', 'CDR3-IMGT', 'JUNCTION', 'J-REGION', 'FR4-IMGT', 'P3V-nt nb', 'N1-REGION-nt nb', 'P5D-nt nb', 'P3D-nt nb', 'N2-REGION-nt nb', 'P5J-nt nb', '3V-REGION trimmed-nt nb', '5D-REGION trimmed-nt nb', '3D-REGION trimmed-nt nb', '5J-REGION trimmed-nt nb'] -vPattern = re.compile(r"IGHV[1-9]-[0-9ab]+-?[1-9]?") -dPattern = re.compile(r"IGHD[1-9]-[0-9ab]+") -jPattern = re.compile(r"IGHJ[1-9]") +#vPattern = re.compile(r"|TRBV[0-9]{1,2}-?[0-9]?") #mouse +#vPattern = re.compile(r"IGHV[1-9]-[0-9ab]+-?[1-9]?") #human +vPattern = re.compile(r"IGHV[1-9]-[0-9ab]+-?[1-9]?|TRBV[0-9]{1,2}-?[0-9]?") #mouse and human + +#dPattern = re.compile(r"TRBD1|TRBD2") #mouse +#dPattern = re.compile(r"IGHD[1-9]-[0-9ab]+") #human +dPattern = re.compile(r"IGHD[1-9]-[0-9ab]+|TRBD1|TRBD2") #mouse and human + + +#jPattern = re.compile(r"TRBJ[12]-[1-7]") #mouse +#jPattern = re.compile(r"IGHJ[1-6]") #human +jPattern = re.compile(r"IGHJ[1-6]|TRBJ[12]-[1-7]") #mouse and human def filterGenes(s, pattern): if type(s) is not str: