def eggnog_map_one(f, cpu=8):
for db_f in ['eggnog.db', 'eggnog_proteins.dmnd']:
if not exists(join('/dev/shm', db_f)):
copy(join(Annotate.OM_RGC.EmapperDir, 'data', db_f),
join('/dev/shm', db_f))
elif os.stat(join('/dev/shm', db_f)).st_size != os.stat(
join(Annotate.OM_RGC.EmapperDir, 'data', db_f)).st_size:
sleep(200) #wait for file to copy
tmpdir = join(Annotate.OM_RGC.AnnotDir, 'tmp', split3way(f)[1])
rmrf(tmpdir)
chdirmkifnotexist(tmpdir)
shell_command("{} {} -i {} --output {} -m diamond --cpu {} --override --data_dir /dev/shm --temp_dir {} --no_file_comment"\
.format(Annotate.OM_RGC.Py27, Annotate.OM_RGC.EmapperPy, f,
join(Annotate.OM_RGC.AnnotDir, split3way(f)[1]), cpu,
tmpdir), verbose=True)
def filt_sort_split(fnames, outfol, out_fname, reference_list, prob_thresh,
threads):
filts = []
unite_f = join(outfol, out_fname + '.u.bam')
sort_f = unite_f.replace('.u.bam', '.s.bam')
if not exists('{}.done'.format(join(outfol, out_fname))):
for fname in fnames:
filt_f = join(outfol,
basename(fname.replace('.icra.bam', '.filt.bam')))
filts.append(filt_f)
do(fname, filt_f, prob_thresh)
samcat(filts, unite_f, delinfile=True)
samsort(unite_f, sort_f, threads, delinfile=True)
shell_command('touch {}.done'.format(join(outfol, out_fname)))
splitbam(sort_f, outfol, reference_list)
def mpilupcall(fnames,
db_fasta,
outvcf,
Q=15,
L=1000,
d=100000,
m=2,
threads=8):
if len(fnames) == 0:
return
cmd = 'bcftools mpileup --threads {} -a FORMAT/AD -Q {} -L {} -d {} -m {} -f {} {}'\
.format(threads, Q, L , d, m, db_fasta, ' '.join(fnames))
cmd += ' | bcftools call --threads {} -Ov -mv -o {}'.format(
threads, outvcf)
shell_command(cmd)
def mpileup_call(dirnm, threads, fasta_db, mpileup_params, del_indir=True):
shell_command('ulimit -n 16384')
outvcf = dirnm + '.vcf'
donef = dirnm + '_calling.done'
if exists(donef):
return
mpilupcall([join(dirnm, '*.bam')],
fasta_db,
outvcf,
Q=mpileup_params['min_base_qual'],
L=mpileup_params['max_depth_indel'],
d=mpileup_params['max_depth'],
m=mpileup_params['min_iReads'],
threads=threads)
if del_indir:
rmrf(dirnm)
shell_command('touch {}'.format(donef))
def samcat(fnames, unite_f, delinfile=True):
shell_command('samtools cat -o {} {}'.format(unite_f, ' '.join(fnames)))
if delinfile:
for fname in fnames:
tryrm(fname)
def samsort(fname_in, fname_out, threads, delinfile=False):
shell_command('samtools sort -@ {} -o {} {}'.format(
threads - 1, fname_out, fname_in))
if delinfile:
tryrm(fname_in)