def build_gene_model(align_db,
clusters,
big_cluster,
find_max,
min_transcript_len=0,
max_isoforms=1e6,
):
'''Build and print out gene models.'''
visited_clusters = set()
transcripts_num = 0
gene_id = 0
excluded = 0
two_exon_trns = set()
def check_criteria(transcript, two_exon_trns):
'''Return True or False whether a transcript pass or
fail the criteria.
'''
transcript_length = sum([align_db.exon_db[e].get_size() \
for e in transcript])
if transcript_length <= min_transcript_len:
return False # fail
else:
if len(transcript) == 2:
trns = ','.join(transcript)
if trns in two_exon_trns:
return False # fail
else:
two_exon_trns.add(trns)
return True # pass
else:
return True
for cl_num, cl in enumerate(big_cluster.nodes(), start=1):
if cl not in visited_clusters:
g = nx.DiGraph()
for intron in clusters[cl].nodes():
g.add_edges_from(align_db.intron_db[intron].edges())
visited_clusters.add(cl)
for neighbor in nx.dfs_tree(big_cluster, cl):
neighbor_cluster = clusters[neighbor]
for intron in neighbor_cluster.nodes():
g.add_edges_from(align_db.intron_db[intron].edges())
visited_clusters.add(neighbor)
if g.nodes():
trans_id = 0
gene_id += 1
collapse_exon(g, align_db)
for node in g.nodes():
if not g.predecessors(node):
g.add_edge('Start', node)
if not g.successors(node):
g.add_edge(node, 'End')
max_paths = [path for path in \
nx.all_simple_paths(g, 'Start', 'End')]
if find_max:
'''Report all maximum isoforms.'''
for transcript in max_paths:
transcript = transcript[1:-1]
if check_criteria(transcript, two_exon_trns):
transcripts_num += 1
trans_id += 1
print_bed(align_db, transcript,
gene_id, trans_id)
else:
excluded += 1
else:
'''Report minimal isoforms if maximum isoforms exceeds
max_isoforms.
'''
if len(max_paths) > max_isoforms:
for transcript in \
get_min_isoforms.get_min_paths(g, False):
if check_criteria(transcript, two_exon_trns):
transcripts_num += 1
trans_id += 1
print_bed(align_db, transcript,
gene_id, trans_id)
else:
excluded += 1
else:
for transcript in max_paths:
transcript = transcript[1:-1]
if check_criteria(transcript, two_exon_trns):
transcripts_num += 1
trans_id += 1
print_bed(align_db, transcript,
gene_id, trans_id)
else:
excluded += 1
if trans_id == 0:
gene_id -= 1
print >> stderr, '\r |--Multi-exon\t\t%d genes, %d isoforms ' % \
(gene_id, transcripts_num),
return gene_id, transcripts_num, excluded
def build_gene_model(genome,
align_db,
clusters,
big_cluster,
find_max,
min_transcript_len=0,
max_isoforms=1e6,
):
'''Build and print out gene models.'''
visited_clusters = set()
transcripts_num = 0
gene_id = 0
excluded = 0
two_exon_trns = set()
def check_criteria(transcript, two_exon_trns):
'''Return True or False whether a transcript pass or
fail the criteria.
'''
transcript_length = sum([align_db.exon_db[e].get_size() \
for e in transcript])
if transcript_length <= min_transcript_len:
return False # fail
else:
if len(transcript) == 2:
trns = ','.join(transcript)
if trns in two_exon_trns:
return False # fail
else:
two_exon_trns.add(trns)
return True # pass
else:
return True
def exon_to_exonobj(exon):
'''Returns an exon objects from a given exon coordinate.'''
chrom, coord = exon.split(':')
start, end = coord.split('-')
return ExonObj(chrom, int(start), int(end))
for cl_num, cl in enumerate(big_cluster.nodes(), start=1):
if cl not in visited_clusters:
g = nx.DiGraph()
for intron in clusters[cl].nodes():
g.add_edges_from(align_db.intron_db[intron].edges())
visited_clusters.add(cl)
for neighbor in nx.dfs_tree(big_cluster, cl):
neighbor_cluster = clusters[neighbor]
for intron in neighbor_cluster.nodes():
g.add_edges_from(align_db.intron_db[intron].edges())
visited_clusters.add(neighbor)
# # nx.draw_spring(nx.algorithms.dfs_tree(g))
# nx.draw_spring(g)
# plt.show()
# for node in g.nodes():
# print node, g[node]
# raise SystemExit
collapse_exon(g, align_db)
for g in split_strand.split(g, genome):
if g.nodes():
subalign_db = AlignmentDB()
for edge in g.edges():
exon1 = exon_to_exonobj(edge[0])
exon2 = exon_to_exonobj(edge[1])
add_exon(subalign_db, [exon1, exon2])
collapse_exon(g, subalign_db)
trans_id = 0
gene_id += 1
strand = g.graph['strand']
for node in g.nodes():
if not g.predecessors(node):
g.add_edge('Start', node)
if not g.successors(node):
g.add_edge(node, 'End')
max_paths = [path for path in \
nx.all_simple_paths(g, 'Start', 'End')]
if find_max:
'''Report all maximum isoforms.'''
for transcript in max_paths:
transcript = transcript[1:-1]
if check_criteria(transcript, two_exon_trns):
transcripts_num += 1
trans_id += 1
print_bed(align_db,
transcript,
strand,
gene_id,
trans_id)
else:
excluded += 1
else:
'''Report minimal isoforms if maximum isoforms exceeds
max_isoforms.
'''
if len(max_paths) > max_isoforms:
for transcript in \
get_min_isoforms.get_min_paths(g, False):
if check_criteria(transcript, two_exon_trns):
transcripts_num += 1
trans_id += 1
print_bed(align_db,
transcript,
strand,
gene_id,
trans_id)
else:
excluded += 1
else:
for transcript in max_paths:
transcript = transcript[1:-1]
if check_criteria(transcript, two_exon_trns):
transcripts_num += 1
trans_id += 1
print_bed(align_db,
transcript,
strand,
gene_id,
trans_id)
else:
excluded += 1
print >> stderr, '\r |--Multi-exon\t\t%d genes, %d isoforms ' % \
(gene_id, transcripts_num),
return gene_id, transcripts_num, excluded
