def phylogeny(args, partition_file, phylip):
basedir = os.getcwd()
try:
os.mkdir('4_phylogeny')
except OSError:
pass
os.chdir('4_phylogeny')
os.symlink(partition_file, partition_file.split('/')[-1])
os.symlink(phylip, phylip.split('/')[-1])
partition_file = partition_file.split('/')[-1]
phylip = phylip.split('/')[-1]
tree = partition_file + '.treefile'
if not os.path.isfile(tree):
job = '{} -nt AUTO -s {} -m MFP -alrt 1000 -bb 1000 -spp {}'
job = job.format(cfg.iqtree, phylip, partition_file)
ID = submit(job,
partition=cfg.SLURMpartition,
account=cfg.SLURMaccount,
qos=cfg.SLURMqos,
time='24:0:0',
job_name='IQTree',
cpus_per_task=cfg.SLURMcpus,
mem_per_cpu=cfg.SLURMmem,
modules=cfg.modules)
job_wait(ID)
err = 'IQTree_{}.err'.format(ID)
out = 'IQTree_{}.out'.format(ID)
cleanup(logs=[out, err])
os.chdir(basedir)
return '4_phylogeny/' + tree
def orthology(args):
# Find orthologous sequences between all genomes and write them to fasta files.
# Align any queries to the reference that have not yet been aligned.
deltafiles = runmummer(args)
OS_file = '1_orthology/' + args.output + '.OS.txt'
if not os.path.isfile(OS_file) or args.force:
job = '{} {} {}'.format(sys.executable, __file__,
args.output + '.args.pickle')
ID = submit(job,
partition=cfg.SLURMpartition,
account=cfg.SLURMaccount,
qos=cfg.SLURMqos,
time='12:0:0',
job_name='py_ortho',
cpus_per_task=cfg.SLURMcpus,
mem_per_cpu=cfg.SLURMmem,
modules=cfg.modules)
job_wait(ID)
out_file = 'py_ortho_' + str(ID) + '.out'
err_file = 'py_ortho_' + str(ID) + '.err'
if not os.path.isfile(OS_file):
exit_message = 'Orthology search failed.\n\n{}:\n{}'
exit_message = exit_message.format(err_file,
open(err_file, 'r').read())
sys.stderr.write(exit_message)
cleanup([out_file, err_file])
return None
else:
cleanup([out_file, err_file])
return get_segments(args)
def submit_tiger2(args, alignment):
# Rate partitioning usually takes under 0.5s per unique pattern
# in the alignment. There is no fast way to know how many unique
# patterns there will be in an alignment of a given size.
pps = 0.1 # empirical guess of high patterns per site # 0.05 for most, 0.1 for extremely diverse clade N. rajui
records = len(args.genomes)
sites = int(total_length(alignment) / records)
estimated_patterns = pps * sites
estimated_runtime = int(estimated_patterns * 0.5)
minutes = int(estimated_runtime / 60) + 1
if minutes > (cfg.LARGEmaxtime * 60):
warning = "Warning: estimated partition time ({}) is greater than maximum wallclock time ({}).\n"
warning = warning.format(minutes, cfg.LARGEmaxtime * 60)
sys.stderr.write(warning)
minutes = cfg.LARGEmaxtime * 60
command = '{} -in {} -a dna -out {} -f phylip -bt rota -b 4 -t 1'
command = command.format(cfg.tiger, alignment, args.output)
ID = submit(command,
partition=cfg.LARGEpartition,
account=cfg.LARGEaccount,
qos=cfg.LARGEqos,
time=str(minutes),
job_name='rate_partitioning',
cpus_per_task=cfg.LARGEcpus,
mem_per_cpu=cfg.LARGEmem,
modules=cfg.modules)
job_wait(ID)
outfile = 'rate_partitioning_' + str(ID) + '.out'
errfile = 'rate_partitioning_' + str(ID) + '.err'
cleanup([outfile, errfile])
def submit_alignment_batch(job):
ID = submit(job,
partition=cfg.LARGEpartition,
account=cfg.LARGEaccount,
qos=cfg.LARGEqos,
time='36:0:0',
job_name='mafft',
cpus_per_task=cfg.LARGEcpus,
mem_per_cpu=cfg.LARGEmem,
modules=cfg.modules)
job_wait(ID)
return ID
def submit_phylogeny(job):
ID = submit(job,
partition = cfg.SLURMpartition,
account = cfg.SLURMaccount,
qos = cfg.SLURMqos,
time = '12:0:0',
job_name = 'ExaML',
cpus_per_task = cfg.SLURMcpus,
mem_per_cpu = cfg.SLURMmem,
modules = cfg.modules)
job_wait(ID)
return ID
def runmummer(args):
basedir = os.getcwd()
try:
os.mkdir('1_orthology')
except OSError:
pass
os.chdir('1_orthology')
existing, absent = find_mummer_files(args)
if absent:
try:
os.mkdir('logs')
except OSError:
pass
jobs = [(nucmer, [args.reference] + absence) for absence in absent]
esttime = '{}:0:0'.format(int(
4 * (len(jobs) / cfg.SLURMcpus + 1))) # estimate 4 hours per query
ID = submit(jobs,
pool=True,
partition=cfg.SLURMpartition,
account=cfg.SLURMaccount,
qos=cfg.SLURMqos,
time=esttime,
job_name='mummer',
cpus_per_task=cfg.SLURMcpus,
mem_per_cpu=cfg.SLURMmem,
modules=cfg.modules)
job_wait(ID)
# all alignments should now exist
existing.extend(absent)
queries, prefixes, deltafiles, tempfiles, filterfiles = list(
zip(*existing))
os.chdir(basedir)
return filterfiles