def junction_exon_triples(chrom, exon_start_stop, transcripts, strand):
from outrigger.index.events import stringify_location
data = []
for transcript, exons in transcripts:
for exon1, exon2 in zip(exons, exons[1:]):
start1, stop1 = exon_start_stop[exon1]
start2, stop2 = exon_start_stop[exon2]
exon1_location = stringify_location(chrom, start1, stop1, strand,
'exon')
exon2_location = stringify_location(chrom, start2, stop2, strand,
'exon')
# if strand == '-':
# start = stop2 + 1
# stop = start1 - 1
# else:
start = stop1 + 1
stop = start2 - 1
junction_location = stringify_location(chrom, start, stop, strand,
'junction')
if strand == '-':
data.append([exon1_location, 'downstream', junction_location])
data.append([exon2_location, 'upstream', junction_location])
else:
data.append([exon1_location, 'upstream', junction_location])
data.append([exon2_location, 'downstream', junction_location])
data = pd.DataFrame(data, columns=['exon', 'direction', 'junction'])
data = data.drop_duplicates()
return data
def graph_items(exon_start_stop, transcripts, chrom, strand):
from outrigger.index.events import stringify_location, opposite
graph = connect(":memory:", graphs=['upstream', 'downstream'])
items = []
triples = set()
for transcript, exons in transcripts:
for exon1, exon2 in zip(exons, exons[1:]):
start1, stop1 = exon_start_stop[exon1]
start2, stop2 = exon_start_stop[exon2]
exon1_location = stringify_location(chrom, start1, stop1, strand,
'exon')
exon2_location = stringify_location(chrom, start2, stop2, strand,
'exon')
start = stop1 + 1
stop = start2 - 1
junction_location = stringify_location(chrom, start, stop, strand,
'junction')
if exon1_location not in items:
items.append(exon1_location)
if exon2_location not in items:
items.append(exon2_location)
if junction_location not in items:
items.append(junction_location)
# Get unique integer for junction
junction_i = items.index(junction_location)
if strand == '-':
exon1_triple = exon1_location, 'downstream', junction_location
exon2_triple = exon2_location, 'upstream', junction_location
else:
exon1_triple = exon1_location, 'upstream', junction_location
exon2_triple = exon2_location, 'downstream', junction_location
exon_triples = exon1_triple, exon2_triple
with graph.transaction() as tr:
for exon_triple in exon_triples:
if exon_triple not in triples:
triples.add(exon_triple)
exon, direction, junction = exon_triple
# Get unique integer for exon
exon_i = items.index(exon)
tr.store(getattr(V(exon_i), direction)(junction_i))
tr.store(
getattr(V(junction_i),
opposite(direction))(exon_i))
else:
continue
items = tuple(items)
return graph, items
def graph_items(exon_start_stop, transcripts, chrom, strand):
from outrigger.index.events import stringify_location, opposite
graph = connect(":memory:", graphs=['upstream', 'downstream'])
items = []
triples = set()
for transcript, exons in transcripts:
for exon1, exon2 in zip(exons, exons[1:]):
start1, stop1 = exon_start_stop[exon1]
start2, stop2 = exon_start_stop[exon2]
exon1_location = stringify_location(chrom, start1, stop1, strand,
'exon')
exon2_location = stringify_location(chrom, start2, stop2, strand,
'exon')
start = stop1 + 1
stop = start2 - 1
junction_location = stringify_location(chrom, start, stop, strand,
'junction')
if exon1_location not in items:
items.append(exon1_location)
if exon2_location not in items:
items.append(exon2_location)
if junction_location not in items:
items.append(junction_location)
# Get unique integer for junction
junction_i = items.index(junction_location)
if strand == '-':
exon1_triple = exon1_location, 'downstream', junction_location
exon2_triple = exon2_location, 'upstream', junction_location
else:
exon1_triple = exon1_location, 'upstream', junction_location
exon2_triple = exon2_location, 'downstream', junction_location
exon_triples = exon1_triple, exon2_triple
with graph.transaction() as tr:
for exon_triple in exon_triples:
if exon_triple not in triples:
triples.add(exon_triple)
exon, direction, junction = exon_triple
# Get unique integer for exon
exon_i = items.index(exon)
tr.store(getattr(V(exon_i), direction)(junction_i))
tr.store(getattr(V(junction_i), opposite(direction))(
exon_i))
else:
continue
items = tuple(items)
return graph, items
def test_stringify_location(chrom, strand, region):
from outrigger.index.events import stringify_location
test = stringify_location(chrom, 100, 200, strand, region)
if region is None:
true = '{0}:{1}-{2}:{3}'.format(chrom, 100, 200, strand)
else:
true = '{0}:{1}:{2}-{3}:{4}'.format(region, chrom, 100, 200, strand)
assert test == true
def test_stringify_location(chrom, strand, region):
from outrigger.index.events import stringify_location
test = stringify_location(chrom, 100, 200, strand, region)
if region is None:
true = '{0}:{1}-{2}:{3}'.format(chrom, 100, 200, strand)
else:
true = '{0}:{1}:{2}-{3}:{4}'.format(region, chrom, 100, 200, strand)
assert test == true
def junction_to_exons(chrom, exon_start_stop, transcripts, strand):
from collections import defaultdict
from outrigger.index.events import stringify_location
data = defaultdict(lambda: {'upstream': set([]), 'downstream': set([])})
for transcript, exons in transcripts:
for exon1, exon2 in zip(exons, exons[1:]):
start1, stop1 = exon_start_stop[exon1]
start2, stop2 = exon_start_stop[exon2]
exon1_location = stringify_location(chrom, start1, stop1, strand,
'exon')
exon2_location = stringify_location(chrom, start2, stop2, strand,
'exon')
# if strand == '-':
# start = stop2 + 1
# stop = start1 - 1
# else:
start = stop1 + 1
stop = start2 - 1
junction_location = stringify_location(chrom, start, stop, strand,
'junction')
if strand == '-':
data[junction_location]['downstream'].add(exon1_location)
data[junction_location]['upstream'].add(exon2_location)
else:
data[junction_location]['upstream'].add(exon1_location)
data[junction_location]['downstream'].add(exon2_location)
data = pd.DataFrame(data).T
data = data.applymap(lambda x: ','.join(x))
data = data.reset_index()
data = data.rename(columns={'index': 'junction'})
return data
def junction_to_exons(chrom, exon_start_stop, transcripts, strand):
from collections import defaultdict
from outrigger.index.events import stringify_location
data = defaultdict(lambda: {'upstream': set([]), 'downstream': set([])})
for transcript, exons in transcripts:
for exon1, exon2 in zip(exons, exons[1:]):
start1, stop1 = exon_start_stop[exon1]
start2, stop2 = exon_start_stop[exon2]
exon1_location = stringify_location(chrom, start1, stop1,
strand, 'exon')
exon2_location = stringify_location(chrom, start2, stop2,
strand, 'exon')
# if strand == '-':
# start = stop2 + 1
# stop = start1 - 1
# else:
start = stop1 + 1
stop = start2 - 1
junction_location = stringify_location(chrom, start,
stop, strand, 'junction')
if strand == '-':
data[junction_location]['downstream'].add(exon1_location)
data[junction_location]['upstream'].add(exon2_location)
else:
data[junction_location]['upstream'].add(exon1_location)
data[junction_location]['downstream'].add(exon2_location)
data = pd.DataFrame(data).T
data = data.applymap(lambda x: ','.join(x))
data = data.reset_index()
data = data.rename(columns={'index': 'junction'})
return data
def junction_exon_triples(chrom, exon_start_stop, transcripts, strand):
from outrigger.index.events import stringify_location
data = []
for transcript, exons in transcripts:
for exon1, exon2 in zip(exons, exons[1:]):
start1, stop1 = exon_start_stop[exon1]
start2, stop2 = exon_start_stop[exon2]
exon1_location = stringify_location(chrom, start1, stop1,
strand, 'exon')
exon2_location = stringify_location(chrom, start2, stop2,
strand, 'exon')
# if strand == '-':
# start = stop2 + 1
# stop = start1 - 1
# else:
start = stop1 + 1
stop = start2 - 1
junction_location = stringify_location(chrom, start, stop,
strand, 'junction')
if strand == '-':
data.append(
[exon1_location, 'downstream', junction_location])
data.append(
[exon2_location, 'upstream', junction_location])
else:
data.append(
[exon1_location, 'upstream', junction_location])
data.append(
[exon2_location, 'downstream', junction_location])
data = pd.DataFrame(data, columns=['exon', 'direction', 'junction'])
data = data.drop_duplicates()
return data
