def get_genes(base, alleles=None):
if alleles is None: # take all of 'em
alleles = [
utils.allele(g) for g in glfo['seqs'][args.region]
if base == get_base(g)
]
return [
args.locus.upper() + args.region.upper() + base + '*' + al
for al in alleles
]
def get_mutfo_from_name(gene_name):
allele_list = utils.allele(gene_name).split("+")
if len(allele_list) != 2:
raise Exception("couldn't get snp info from gene name %s" % gene_name)
mutfo = {}
for mutstr in allele_list[1].split("."):
if len(mutstr) < 3:
raise Exception("couldn't extract mutation info from %s" % mutstr)
original, new = mutstr[0], mutstr[-1]
if original not in utils.nukes or new not in utils.nukes:
raise Exception("couldn't extract mutation info from %s" % mutstr)
position = int(mutstr[1:-1])
if position not in mutfo:
mutfo[position] = {}
mutfo[position]["original"] = original # if it *is* already there, we want to *keep* the old 'original'
mutfo[position]["new"] = new
if mutfo[position]["new"] == mutfo[position]["original"]: # reverted back to the original base
del mutfo[position]
return mutfo
def get_mutfo_from_name(gene_name):
allele_list = utils.allele(gene_name).split('+')
if len(allele_list) != 2:
raise Exception('couldn\'t get snp info from gene name %s' % gene_name)
mutfo = {}
for mutstr in allele_list[1].split('.'):
if len(mutstr) < 3:
raise Exception('couldn\'t extract mutation info from %s' % mutstr)
original, new = mutstr[0], mutstr[-1]
if original not in utils.nukes or new not in utils.nukes:
raise Exception('couldn\'t extract mutation info from %s' % mutstr)
position = int(mutstr[1:-1])
if position not in mutfo:
mutfo[position] = {}
mutfo[position]['original'] = original # if it *is* already there, we want to *keep* the old 'original'
mutfo[position]['new'] = new
if mutfo[position]['new'] == mutfo[position]['original']: # reverted back to the original base
del mutfo[position]
return mutfo
def get_new_allele_name_and_change_mutfo(template_gene, mutfo):
if '+' in utils.allele(template_gene): # template gene was already snp'd
old_mutfo = get_mutfo_from_name(template_gene)
for position, info in mutfo.items():
if position not in old_mutfo:
old_mutfo[position] = {}
old_mutfo[position]['original'] = info['original'] # if it *is* already there, we want to *keep* the old 'original'
old_mutfo[position]['new'] = info['new']
if old_mutfo[position]['new'] == old_mutfo[position]['original']: # reverted back to the original base
del old_mutfo[position]
final_mutfo = old_mutfo
assert len(template_gene.split('+')) == 2
template_gene = template_gene.split('+')[0] # before we did any snp'ing
else:
final_mutfo = mutfo
final_name = template_gene
if len(final_mutfo) > 0:
assert '+' not in final_name
final_name += '+' + stringify_mutfo(final_mutfo) # full, but possibly overly verbose
return final_name, final_mutfo
def get_new_allele_name_and_change_mutfo(template_gene, mutfo):
if '+' in utils.allele(template_gene): # template gene was already snp'd
old_mutfo = get_mutfo_from_name(template_gene)
for position, info in mutfo.items():
if position not in old_mutfo:
old_mutfo[position] = {}
old_mutfo[position]['original'] = info['original'] # if it *is* already there, we want to *keep* the old 'original'
old_mutfo[position]['new'] = info['new']
if old_mutfo[position]['new'] == old_mutfo[position]['original']: # reverted back to the original base
del old_mutfo[position]
final_mutfo = old_mutfo
assert len(template_gene.split('+')) == 2
template_gene = template_gene.split('+')[0] # before we did any snp'ing
else:
final_mutfo = mutfo
final_name = template_gene
if len(final_mutfo) > 0:
assert '+' not in final_name
final_name += '+' + stringify_mutfo(final_mutfo) # full, but possibly overly verbose
return final_name, final_mutfo
sys.path.insert(1, partis_dir + '/python')
import utils
import glutils
parser = argparse.ArgumentParser()
parser.add_argument('--base', required=True)
parser.add_argument('--alleles')
parser.add_argument('--other-genes')
parser.add_argument('--region', default='v')
parser.add_argument('--chain', default='h')
parser.add_argument('--glfo-dir', default='data/germlines/human')
args = parser.parse_args()
glfo = glutils.read_glfo(args.glfo_dir, args.chain)
if args.alleles is None:
args.alleles = [utils.allele(g) for g in glfo['seqs'][args.region] if args.base == utils.primary_version(g) + '-' + utils.sub_version(g)]
else:
args.alleles = utils.get_arg_list(args.alleles)
args.other_genes = utils.get_arg_list(args.other_genes)
# for g, s in glfo['seqs']['v'].items():
# print '%s %3d' % (utils.color_gene(g, width=20), len(s) - glfo['cyst-positions'][g])
# sys.exit()
# base = '4-59'
# a1, a2 = '12', '01'
# gene1, gene2 = 'IGHV' + base + '*' + a1, 'IGHV' + base + '*' + a2
genes = ['IG' + args.chain.upper() + args.region.upper() + args.base + '*' + al for al in args.alleles]
if args.other_genes is not None:
genes += args.other_genes
import utils
import glutils
parser = argparse.ArgumentParser()
parser.add_argument('--base', required=True)
parser.add_argument('--alleles')
parser.add_argument('--other-genes')
parser.add_argument('--region', default='v')
parser.add_argument('--locus', default='igh', choices=utils.loci.keys())
parser.add_argument('--glfo-dir', default='data/germlines/human')
args = parser.parse_args()
glfo = glutils.read_glfo(args.glfo_dir, args.locus)
if args.alleles is None:
args.alleles = [
utils.allele(g) for g in glfo['seqs'][args.region]
if args.base == utils.primary_version(g) +
('-' +
utils.sub_version(g) if utils.sub_version(g) is not None else '')
]
else:
args.alleles = utils.get_arg_list(args.alleles)
if len(args.alleles) == 0:
raise Exception(
'couldn\'t find any alleles for --base %s. Other choices:\n %s' %
(args.base, ' '.join(glfo['seqs'][args.region])))
args.other_genes = utils.get_arg_list(args.other_genes)
# for g, s in glfo['seqs']['v'].items():
# print '%s %3d' % (utils.color_gene(g, width=20), len(s) - glfo['cyst-positions'][g])
# sys.exit()
