def execute_megablast(self, fasta1, fasta2):
import os
from TPutils import shell_command
for one_fasta in fasta2:
out_prefix = os.path.basename(one_fasta).split('.')[0]
cmd1 = "formatdb -i %s -p F" % one_fasta
cmd2 = 'blastn -task megablast -query %s -db %s -evalue 1e-5 -outfmt 6 -out blast_result_%s.tab' % (
fasta1, one_fasta, out_prefix)
a, b, c = shell_command.shell_command(cmd1)
a, b, c = shell_command.shell_command(cmd2)
def delta_file2start_stop_list(delta_input,
contigs_add,
algo='nucmer',
minimum_identity=85):
'''
:param coords_input:
:param contig_add:
:param algo:
:return:
'''
import re
from TPutils import shell_command
out_name = delta_input.split('.')[0]
coords_file = 'gaps_%s.coords' % out_name
cmd2 = 'show-coords -T -r -c -L 100 -I %s %s > %s' % (
minimum_identity, delta_input, coords_file)
print(cmd2)
a, b, c = shell_command.shell_command(cmd2)
if c != 0:
print('nucmer error!!!')
print(a, b, c)
print(cmd2)
import sys
sys.exit()
with open("gaps_%s.coords" % out_name, 'rU') as infile:
rawLinks = justLinks(infile.readlines())
if algo == 'promer':
shift = 4
elif algo == 'nucmer':
shift = 0
contig2start_stop_list = {}
i = 1
for n, row in enumerate(rawLinks):
l = re.split(r'\t+', row.rstrip('\n'))
start = int(l[0]) + contigs_add[l[9 + shift]][0]
stop = int(l[1]) + contigs_add[l[9 + shift]][0]
contig_ref = l[9 + shift]
if contig_ref not in contig2start_stop_list:
contig2start_stop_list[contig_ref] = {}
contig2start_stop_list[contig_ref]["start"] = [start]
contig2start_stop_list[contig_ref]["stop"] = [stop]
else:
contig2start_stop_list[contig_ref]["start"].append(start)
contig2start_stop_list[contig_ref]["stop"].append(stop)
return contig2start_stop_list, coords_file
def run_circos(self, config_file="circos.config", out_prefix="circos"):
from TPutils import shell_command
cmd = 'circos -outputfile %s.svg -conf %s' % (out_prefix, config_file)
a, b, c = shell_command.shell_command(cmd)
sys.stdout.write(str(a))
sys.stdout.write(str(b))
sys.stdout.write(str(c))
sys.stdout.flush()
if c == 255:
raise CircosException("Circos problem, check files... quitting")
def blast2circos_file(self,
blast,
reference,
blastn=False,
identity_cutoff=80):
'''
tblastn vs contigs by default
can be switch to blastn
:param blast:
:param reference:
:param blastn:
:return:
'''
from TPutils import shell_command
from TPutils import blast_utils
from Bio.Blast.Applications import NcbitblastnCommandline
from Bio.Blast.Applications import NcbiblastnCommandline
# todo catch IO errors, orther potential errors
a, b, c = shell_command.shell_command('formatdb -i %s -p F' %
(reference))
# print a
# print b
print(c)
if not blastn:
blast_cline = NcbitblastnCommandline(
query=blast,
db=reference,
evalue=0.00000001, # 0.001
outfmt=6,
out="blast.tmp",
max_target_seqs=1)
print(blast_cline)
else:
blast_cline = NcbiblastnCommandline(query=blast,
db=reference,
evalue=0.001,
outfmt=6,
out="blast.tmp")
stdout, stderr = blast_cline()
# a,b,c = shell_command.shell_command('tblastn -query %s -db %s -evalue 1e-5 -max_target_seqs 1 -outfmt 6 > blast.tmp' % (blast, reference))
# a,b,c = shell_command.shell_command('tblastn -query %s -db %s -evalue 1e-5 -max_target_seqs 1 -outfmt 6' % (blast, reference))
print('############## BLAST ###################')
# print a
# print b
# print c
blast2data, queries = blast_utils.remove_blast_redundancy(
["blast.tmp"], check_overlap=False)
o = open('circos_blast.txt', "w")
l = open('circos_blast_labels.txt', "w")
# with open(blast, 'r') as b:
'''
for line in a.split('\n'):
data = line.rstrip().split('\t')
#print data
try:
if float(data[2])>80:
location = sorted([data[8], data[9]])
o.write("%s\t%s\t%s\n" % (data[1], location[0], location[1]))
l.write("%s\t%s\t%s\t%s\n" % (data[1], location[0], location[1], data[0]))
except IndexError:
continue
'''
for contig in blast2data:
cname = re.sub("\|", "", contig)
for gene in blast2data[contig]:
if float(blast2data[contig][gene]
[0]) >= identity_cutoff: # 80,20
location = sorted(blast2data[contig][gene][1:3])
o.write("%s\t%s\t%s\n" %
(contig, location[0] + self.contigs_add[cname][0],
location[1] + self.contigs_add[cname][0]))
l.write("%s\t%s\t%s\t%s\n" %
(contig, location[0] + self.contigs_add[cname][0],
location[1] + self.contigs_add[cname][0], gene))
o.close()
def execute_promer(fasta1,
fasta2,
algo="nucmer",
coords=True,
minimum_align_length=100,
minimum_identity=30):
'''
:param fasta1: reference fata
:param fasta2: query fasta(s)
:param algo: nucmer/promer
:param coords: execute show-coords
:param minimum_align_length: default:100
:param minimum_identity: default:30
:return: coords file list, delta file list
'''
from TPutils import shell_command
import os
#cmd1 = 'promer --mum -l 5 %s %s' % (fasta1, fasta2)
delta_files = []
coord_files = []
for one_fasta in fasta2:
if algo == 'nucmer':
# 03.2017 => changed from -mum to -mumreference. 01.06.17 removed -mumreference
cmd1 = 'nucmer -b 200 -c 65 -g 90 -l 20 -p %s %s %s' % (
os.path.basename(one_fasta).split('.')[0], fasta1, one_fasta)
print(cmd1)
a, b, c = shell_command.shell_command(cmd1)
if c != 0:
raise Exception("%s" % b)
if coords:
cmd2 = 'show-coords -T -r -c -L %s -I %s %s.delta > %s.coords' % (
minimum_align_length, minimum_identity,
os.path.basename(one_fasta).split('.')[0],
os.path.basename(one_fasta).split('.')[0])
print(cmd2)
a, b, c = shell_command.shell_command(cmd2)
if c != 0:
raise Exception("%s" % b)
elif algo == 'promer':
# promer --mum -l 5
cmd1 = 'promer -l 5 -p %s %s %s' % (
os.path.basename(one_fasta).split('.')[0], fasta1, one_fasta)
a, b, c = shell_command.shell_command(cmd1)
if c != 0:
raise Exception("%s" % b)
if coords:
# show-coords -T -r -c -L 100 -I 30 out.delta
cmd2 = 'show-coords -T -r -c -L %s -I %s %s.delta > %s.coords' % (
minimum_align_length, minimum_identity,
os.path.basename(one_fasta).split('.')[0],
os.path.basename(one_fasta).split('.')[0])
print(cmd2)
a, b, c = shell_command.shell_command(cmd2)
if c != 0:
raise Exception("%s" % b)
delta_files.append('%s.delta' %
os.path.basename(one_fasta).split('.')[0])
if coords:
coord_files.append('%s.coords' %
os.path.basename(one_fasta).split('.')[0])
return coord_files, delta_files
