def glimmerHMM(tool_dir, fasta_fpath, out_fpath, gene_lengths, err_path, tmp_dir, index):
def run(contig_path, tmp_path):
with open(err_path, 'a') as err_file:
return_code = qutils.call_subprocess(
[tool_exec, contig_path, '-d', trained_dir, '-g', '-o', tmp_path],
stdout=err_file,
stderr=err_file,
indent=' ' + qutils.index_to_str(index) + ' ')
return return_code
tool_exec = os.path.join(tool_dir, 'glimmerhmm')
# Note: why arabidopsis? for no particular reason, really.
trained_dir = os.path.join(tool_dir, 'trained', 'arabidopsis')
contigs = {}
gffs = []
base_dir = tempfile.mkdtemp(dir=tmp_dir)
for seq_num, (ind, seq) in enumerate(read_fasta(fasta_fpath)):
seq_num = str(seq_num)
ind = ind[:qutils.MAX_CONTIG_NAME_GLIMMER]
contig_path = os.path.join(base_dir, seq_num + '.fasta')
gff_path = os.path.join(base_dir, seq_num + '.gff')
write_fasta(contig_path, [(ind, seq)])
if run(contig_path, gff_path) == 0:
gffs.append(gff_path)
contigs[ind] = seq
if not gffs:
return None, None, None, None
out_gff_fpath = out_fpath + '_genes.gff' + ('.gz' if not qconfig.no_gzip else '')
out_gff_path = merge_gffs(gffs, out_gff_fpath)
unique, total = set(), 0
genes = []
for contig, gene_id, start, end, strand in parse_gff(out_gff_path):
total += 1
if strand == '+':
gene_seq = contigs[contig][start - 1:end]
else:
gene_seq = rev_comp(contigs[contig][start - 1:end])
if gene_seq not in unique:
unique.add(gene_seq)
gene = Gene(contig=contig, start=start, end=end, strand=strand, seq=gene_seq)
gene.is_full = gene.start > 1 and gene.end < len(contigs[contig])
genes.append(gene)
full_cnt = [sum([gene.end - gene.start >= threshold for gene in genes if gene.is_full]) for threshold in gene_lengths]
partial_cnt = [sum([gene.end - gene.start >= threshold for gene in genes if not gene.is_full]) for threshold in gene_lengths]
if OUTPUT_FASTA:
out_fasta_fpath = out_fpath + '_genes.fasta'
add_genes_to_fasta(genes, out_fasta_fpath)
if not qconfig.debug:
shutil.rmtree(base_dir)
#return out_gff_path, out_fasta_path, len(unique), total, cnt
return out_gff_path, genes, len(unique), total, full_cnt, partial_cnt
def glimmerHMM(tool_dir, fasta_fpath, out_fpath, gene_lengths, err_path, tmp_dir, index):
def run(contig_path, tmp_path):
with open(err_path, 'a') as err_file:
return_code = qutils.call_subprocess(
[tool_exec, contig_path, '-d', trained_dir, '-g', '-o', tmp_path],
stdout=err_file,
stderr=err_file,
indent=' ' + qutils.index_to_str(index) + ' ')
return return_code
tool_exec = os.path.join(tool_dir, 'glimmerhmm')
# Note: why arabidopsis? for no particular reason, really.
trained_dir = os.path.join(tool_dir, 'trained', 'arabidopsis')
contigs = {}
gffs = []
base_dir = tempfile.mkdtemp(dir=tmp_dir)
for seq_num, (ind, seq) in enumerate(read_fasta(fasta_fpath)):
seq_num = str(seq_num)
ind = ind[:qutils.MAX_CONTIG_NAME_GLIMMER]
contig_path = os.path.join(base_dir, seq_num + '.fasta')
gff_path = os.path.join(base_dir, seq_num + '.gff')
write_fasta(contig_path, [(ind, seq)])
if run(contig_path, gff_path) == 0:
gffs.append(gff_path)
contigs[ind] = seq
if not gffs:
return None, None, None, None
out_gff_fpath = out_fpath + '_genes.gff' + ('.gz' if not qconfig.no_gzip else '')
out_gff_path = merge_gffs(gffs, out_gff_fpath)
unique, total = set(), 0
genes = []
for contig, gene_id, start, end, strand in parse_gff(out_gff_path):
total += 1
if strand == '+':
gene_seq = contigs[contig][start - 1:end]
else:
gene_seq = rev_comp(contigs[contig][start - 1:end])
if gene_seq not in unique:
unique.add(gene_seq)
gene = Gene(contig=contig, start=start, end=end, strand=strand, seq=gene_seq)
gene.is_full = gene.start > 1 and gene.end < len(contigs[contig])
genes.append(gene)
full_cnt = [sum([gene.end - gene.start >= threshold for gene in genes if gene.is_full]) for threshold in gene_lengths]
partial_cnt = [sum([gene.end - gene.start >= threshold for gene in genes if not gene.is_full]) for threshold in gene_lengths]
if OUTPUT_FASTA:
out_fasta_fpath = out_fpath + '_genes.fasta'
add_genes_to_fasta(genes, out_fasta_fpath)
if not qconfig.debug:
shutil.rmtree(base_dir)
#return out_gff_path, out_fasta_path, len(unique), total, cnt
return out_gff_path, genes, len(unique), total, full_cnt, partial_cnt
def parse_gmhmm_out(out_fpath):
reading_gene = False
reading_protein = False
protein = ''
with open(out_fpath) as f:
for line in f:
if line.startswith('>gene'):
seq = []
seq_id, contig_id = line.strip().split('\t')
# >gene_2|GeneMark.hmm|57_nt|+|1|57 >NODE_3_length_713_cov_1.25228
_, _, seq_len, strand, left_index, right_index = seq_id.split('|')
contig_id = contig_id[1:]
if 'nt' in seq_len:
reading_gene = True
elif 'aa' in seq_len:
reading_protein = True
elif reading_gene or reading_protein:
if line.isspace():
left_index = int(left_index)
right_index = int(right_index)
if reading_gene:
seq = ''.join(seq)
reading_gene = False
elif reading_protein:
protein = ''.join(seq)
seq = []
reading_protein = False
#genes.append(Gene(contig_id, strand, left_index, right_index, str_seq))
if seq:
gene = Gene(contig=contig_id, start=left_index, end=right_index, strand=strand, seq=seq, protein=protein)
yield gene
else:
seq.append(line.strip())
def parse_gtf_out(out_fpath):
with open(out_fpath) as f:
for line in f:
if 'CDS' in line:
l = line.strip().split()
gene = Gene(contig=l[0], strand=l[6], start=int(l[3]), end=int(l[4]), seq=l[9])
yield gene
def parse_gmhmm_out(out_fpath):
reading_gene = False
reading_protein = False
protein = ''
genes_by_id = OrderedDict()
gene_id = None
with open(out_fpath) as f:
for line in f:
if line.startswith('>gene'):
seq = []
seq_id, contig_id = line.strip().split('\t')
# >gene_2|GeneMark.hmm|57_nt|+|1|57 >NODE_3_length_713_cov_1.25228
gene_id, _, seq_len, strand, left_index, right_index = seq_id.split(
'|')
gene_id = gene_id[1:]
contig_id = contig_id[1:]
if 'nt' in seq_len:
reading_gene = True
elif 'aa' in seq_len:
reading_protein = True
elif reading_gene or reading_protein:
if line.isspace():
left_index = int(left_index)
right_index = int(right_index)
if reading_gene:
seq = ''.join(seq)
reading_gene = False
elif reading_protein:
protein = ''.join(seq)
seq = []
reading_protein = False
#genes.append(Gene(contig_id, strand, left_index, right_index, str_seq))
gene = genes_by_id[gene_id] if gene_id in genes_by_id else \
Gene(contig=contig_id, start=left_index, end=right_index, strand=strand)
if seq:
gene.seq = seq
seq = []
if protein:
gene.protein = protein
protein = None
genes_by_id[gene_id] = gene
else:
seq.append(line.strip())
return list(genes_by_id.values())