def test_split_fasta_diff_num_seqs_per_file_alt(self):
"""split_fasta funcs always catches all seqs
"""
# start with 59 seqs (b/c it's prime, so should make more
# confusing splits)
in_seqs = LoadSeqs(data=[('seq%s' % k,'AACCTTAA') for k in range(59)])
infile = in_seqs.toFasta().split('\n')
# test seqs_per_file from 1 to 1000
for i in range(1,1000):
filename_prefix = get_random_job_prefix(fixed_prefix='/tmp/')
actual = split_fasta(infile, i, filename_prefix)
actual_seqs = []
for fp in actual:
actual_seqs += list(open(fp))
# remove the files now, so if the test fails they still get
# cleaned up
remove_files(actual)
# building seq collections from infile and the split files result in
# equivalent seq collections
self.assertEqual(\
LoadSeqs(data=infile,aligned=False),\
LoadSeqs(data=actual_seqs,aligned=False))
def test_split_fasta_diff_num_seqs_per_file_alt(self):
"""split_fasta funcs always catches all seqs
"""
# start with 59 seqs (b/c it's prime, so should make more
# confusing splits)
in_seqs = LoadSeqs(data=[('seq%s' % k, 'AACCTTAA') for k in range(59)])
infile = in_seqs.toFasta().split('\n')
# test seqs_per_file from 1 to 1000
for i in range(1, 1000):
filename_prefix = get_random_job_prefix(fixed_prefix='/tmp/')
actual = split_fasta(infile, i, filename_prefix)
actual_seqs = []
for fp in actual:
actual_seqs += list(open(fp))
# remove the files now, so if the test fails they still get
# cleaned up
remove_files(actual)
# building seq collections from infile and the split files result in
# equivalent seq collections
self.assertEqual(\
LoadSeqs(data=infile,aligned=False),\
LoadSeqs(data=actual_seqs,aligned=False))
def test_get_random_job_prefix(self):
""" get_random_job_prefix functions as expected """
s1 = get_random_job_prefix()
s2 = get_random_job_prefix()
self.assertNotEqual(s1, s2)
self.assertEqual(len(s1), 10)
self.assertEqual(len(s2), 10)
# different max len
s1 = get_random_job_prefix(max_job_prefix_len=22)
self.assertEqual(len(s1), 22)
# fixed_prefix added
s1 = get_random_job_prefix(fixed_prefix='TEST')
s2 = get_random_job_prefix(fixed_prefix='TEST')
self.assertNotEqual(s1, s2)
self.assertEqual(len(s1), 10)
self.assertTrue(s1.startswith('TEST'))
self.assertTrue(s2.startswith('TEST'))
# leading/trailing underscores added
self.assertTrue(s1.startswith('TEST_'))
self.assertTrue(s1.endswith('_'))
# no leading/trailing underscores
s1 = get_random_job_prefix(leading_trailing_underscores=False)
self.assertFalse(s1.startswith('_'))
self.assertFalse(s1.endswith('_'))
# combo of all parameters
s1 = get_random_job_prefix(leading_trailing_underscores=False,\
fixed_prefix='HELLO',max_job_prefix_len=12)
self.assertEqual(len(s1), 12)
self.assertTrue(s1.startswith('HELLO'))
self.assertFalse(s1.endswith('_'))
def test_get_random_job_prefix(self):
""" get_random_job_prefix functions as expected """
s1 = get_random_job_prefix()
s2 = get_random_job_prefix()
self.assertNotEqual(s1,s2)
self.assertEqual(len(s1),10)
self.assertEqual(len(s2),10)
# different max len
s1 = get_random_job_prefix(max_job_prefix_len=22)
self.assertEqual(len(s1),22)
# fixed_prefix added
s1 = get_random_job_prefix(fixed_prefix='TEST')
s2 = get_random_job_prefix(fixed_prefix='TEST')
self.assertNotEqual(s1,s2)
self.assertEqual(len(s1),10)
self.assertTrue(s1.startswith('TEST'))
self.assertTrue(s2.startswith('TEST'))
# leading/trailing underscores added
self.assertTrue(s1.startswith('TEST_'))
self.assertTrue(s1.endswith('_'))
# no leading/trailing underscores
s1 = get_random_job_prefix(leading_trailing_underscores=False)
self.assertFalse(s1.startswith('_'))
self.assertFalse(s1.endswith('_'))
# combo of all parameters
s1 = get_random_job_prefix(leading_trailing_underscores=False,\
fixed_prefix='HELLO',max_job_prefix_len=12)
self.assertEqual(len(s1),12)
self.assertTrue(s1.startswith('HELLO'))
self.assertFalse(s1.endswith('_'))
def test_split_fasta_equal_num_seqs_per_file(self):
"""split_fasta funcs as expected when equal num seqs go to each file
"""
filename_prefix = get_random_job_prefix(fixed_prefix='/tmp/')
infile = ['>seq1','AACCTTAA','>seq2','TTAACC','AATTAA',\
'>seq3','CCTT--AA']
actual = split_fasta(infile, 1, filename_prefix)
actual_seqs = []
for fp in actual:
actual_seqs += list(open(fp))
remove_files(actual)
expected = ['%s.%d.fasta' % (filename_prefix, i) for i in range(3)]
self.assertEqual(actual, expected)
self.assertEqual(\
LoadSeqs(data=infile,aligned=False),\
LoadSeqs(data=actual_seqs,aligned=False))
def test_split_fasta_equal_num_seqs_per_file(self):
"""split_fasta funcs as expected when equal num seqs go to each file
"""
filename_prefix = get_random_job_prefix(fixed_prefix='/tmp/')
infile = ['>seq1','AACCTTAA','>seq2','TTAACC','AATTAA',\
'>seq3','CCTT--AA']
actual = split_fasta(infile, 1, filename_prefix)
actual_seqs = []
for fp in actual:
actual_seqs += list(open(fp))
remove_files(actual)
expected = ['%s.%d.fasta' % (filename_prefix,i) for i in range(3)]
self.assertEqual(actual,expected)
self.assertEqual(\
LoadSeqs(data=infile,aligned=False),\
LoadSeqs(data=actual_seqs,aligned=False))
def test_split_fasta_diff_num_seqs_per_file(self):
"""split_fasta funcs as expected when diff num seqs go to each file
"""
filename_prefix = get_random_job_prefix(fixed_prefix='/tmp/')
infile = ['>seq1','AACCTTAA','>seq2','TTAACC','AATTAA',\
'>seq3','CCTT--AA']
actual = split_fasta(infile, 2, filename_prefix)
actual_seqs = []
for fp in actual:
actual_seqs += list(open(fp))
remove_files(actual)
expected = ['%s.%d.fasta' % (filename_prefix, i) for i in range(2)]
# list of file paths is as expected
self.assertEqual(actual, expected)
# building seq collections from infile and the split files result in
# equivalent seq collections
self.assertEqual(\
LoadSeqs(data=infile,aligned=False),\
LoadSeqs(data=actual_seqs,aligned=False))
def test_split_fasta_diff_num_seqs_per_file(self):
"""split_fasta funcs as expected when diff num seqs go to each file
"""
filename_prefix = get_random_job_prefix(fixed_prefix='/tmp/')
infile = ['>seq1','AACCTTAA','>seq2','TTAACC','AATTAA',\
'>seq3','CCTT--AA']
actual = split_fasta(infile, 2, filename_prefix)
actual_seqs = []
for fp in actual:
actual_seqs += list(open(fp))
remove_files(actual)
expected = ['%s.%d.fasta' % (filename_prefix,i) for i in range(2)]
# list of file paths is as expected
self.assertEqual(actual,expected)
# building seq collections from infile and the split files result in
# equivalent seq collections
self.assertEqual(\
LoadSeqs(data=infile,aligned=False),\
LoadSeqs(data=actual_seqs,aligned=False))
def main():
option_parser, opts, args = parse_command_line_parameters(**script_info)
if opts.blast_db == None and opts.refseqs_fp == None:
option_parser.error('Either blast_db or refseqs_fp must be provided.')
# create local copies of command-line options
python_exe_fp = opts.python_exe_fp
pick_otus_fp = opts.pick_otus_fp
refseqs_fp = opts.refseqs_fp
cluster_jobs_fp = opts.cluster_jobs_fp
input_fasta_fp = opts.input_fasta_fp
jobs_to_start = opts.jobs_to_start
output_dir = opts.output_dir
poller_fp = opts.poller_fp
retain_temp_files = opts.retain_temp_files
suppress_polling = opts.suppress_polling
seconds_to_sleep = opts.seconds_to_sleep
max_e_value = opts.max_e_value
similarity = opts.similarity
poll_directly = opts.poll_directly
min_aligned_percent = opts.min_aligned_percent
created_temp_paths = []
if not opts.blast_db:
# Build the blast database from the reference_seqs_fp -- all procs
# will then access one db rather than create one per proc
blast_db, db_files_to_remove = \
build_blast_db_from_fasta_path(refseqs_fp)
created_temp_paths += db_files_to_remove
else:
blast_db = opts.blast_db
# split the input filepath into directory and filename, base filename and
# extension
input_dir, input_fasta_fn = split(input_fasta_fp)
input_file_basename, input_fasta_ext = splitext(input_fasta_fn)
# set the job_prefix either based on what the user passed in,
# or a random string beginning with RDP
job_prefix = opts.job_prefix or get_random_job_prefix('POTU')
# A temporary output directory is created in output_dir named
# job_prefix. Output files are then moved from the temporary
# directory to the output directory when they are complete, allowing
# a poller to detect when runs complete by the presence of their
# output files.
working_dir = '%s/%s' % (output_dir,job_prefix)
try:
makedirs(working_dir)
created_temp_paths.append(working_dir)
except OSError:
# working dir already exists
pass
# compute the number of sequences that should be included in
# each file after splitting the input fasta file
num_seqs_per_file = compute_seqs_per_file(input_fasta_fp,jobs_to_start)
# split the fasta files and get the list of resulting files
tmp_fasta_fps =\
split_fasta(open(input_fasta_fp),num_seqs_per_file,\
job_prefix,working_dir=output_dir)
created_temp_paths += tmp_fasta_fps
# build the filepath for the 'jobs script'
jobs_fp = '%s/%sjobs.txt' % (output_dir, job_prefix)
created_temp_paths.append(jobs_fp)
# generate the list of commands to be pushed out to nodes and the list of
# output files generated by each job
commands, job_result_filepaths = \
get_job_commands(python_exe_fp,pick_otus_fp,tmp_fasta_fps,
output_dir,blast_db,job_prefix,working_dir,max_e_value,similarity,
min_aligned_percent)
created_temp_paths += job_result_filepaths
# Set up poller apparatus if the user does not suppress polling
if not suppress_polling:
# Write the list of files which must exist for the jobs to be
# considered complete
expected_files_filepath = '%s/expected_out_files.txt' % working_dir
write_filepaths_to_file(job_result_filepaths,expected_files_filepath)
created_temp_paths.append(expected_files_filepath)
# Write the mapping file which described how the output files from
# each job should be merged into the final output files
merge_map_filepath = '%s/merge_map.txt' % working_dir
process_run_results_f =\
'qiime.parallel.pick_otus_blast.parallel_blast_process_run_results_f'
write_merge_map_file_pick_otus(job_result_filepaths,output_dir,\
merge_map_filepath,input_file_basename)
created_temp_paths.append(merge_map_filepath)
# Create the filepath listing the temporary files to be deleted,
# but don't write it yet
deletion_list_filepath = '%s/deletion_list.txt' % working_dir
created_temp_paths.append(deletion_list_filepath)
# Generate the command to run the poller, and the list of temp files
# created by the poller
if not poll_directly:
poller_command, poller_result_filepaths =\
get_poller_command(python_exe_fp,poller_fp,expected_files_filepath,\
merge_map_filepath,deletion_list_filepath,process_run_results_f,\
seconds_to_sleep=seconds_to_sleep)
created_temp_paths += poller_result_filepaths
# append the poller command to the list of job commands
commands.append(poller_command)
else:
poller_command, poller_result_filepaths =\
get_poller_command(python_exe_fp,poller_fp,expected_files_filepath,\
merge_map_filepath,deletion_list_filepath,process_run_results_f,\
seconds_to_sleep=seconds_to_sleep,command_prefix='',command_suffix='')
created_temp_paths += poller_result_filepaths
if not retain_temp_files:
# If the user wants temp files deleted, now write the list of
# temp files to be deleted
write_filepaths_to_file(created_temp_paths,deletion_list_filepath)
else:
# Otherwise just write an empty file
write_filepaths_to_file([],deletion_list_filepath)
# write the commands to the 'jobs files'
write_jobs_file(commands,job_prefix=job_prefix,jobs_fp=jobs_fp)
# submit the jobs file using cluster_jobs, if not suppressed by the
# user
if not opts.suppress_submit_jobs:
submit_jobs(cluster_jobs_fp,jobs_fp,job_prefix)
if poll_directly:
try:
check_call(poller_command.split())
except CalledProcessError, e:
print '**Error occuring when calling the poller directly. '+\
'Jobs may have been submitted, but are not being polled.'
print str(e)
exit(-1)
def main():
option_parser, opts, args = parse_command_line_parameters(**script_info)
# create local copies of command-line options
input_path = opts.input_path
output_dir = opts.output_path
min_seqs = opts.min
max_seqs = opts.max
step = opts.step
num_reps = opts.num_reps
lineages_included = opts.lineages_included
single_rarefaction_fp = opts.single_rarefaction_fp
python_exe_fp = opts.python_exe_fp
path_to_cluster_jobs = opts.cluster_jobs_fp
poller_fp = opts.poller_fp
retain_temp_files = opts.retain_temp_files
suppress_polling = opts.suppress_polling
seconds_to_sleep = opts.seconds_to_sleep
poll_directly = opts.poll_directly
jobs_to_start = opts.jobs_to_start
created_temp_paths = []
# split the input filepath into directory and filename, base filename and
# extension
input_dir, input_fn = split(input_path)
input_file_basename, input_file_ext = splitext(input_fn)
# set the job_prefix either based on what the user passed in,
# or a random string beginning with ALDIV (ALphaDIVersity)
job_prefix = opts.job_prefix or get_random_job_prefix('RARIF')
# A temporary output directory is created in output_dir named
# job_prefix. Output files are then moved from the temporary
# directory to the output directory when they are complete, allowing
# a poller to detect when runs complete by the presence of their
# output files.
working_dir = '%s/%s' % (output_dir,job_prefix)
try:
makedirs(working_dir)
created_temp_paths.append(working_dir)
except OSError:
# working_dir already exists
pass
# build the filepath for the 'jobs script'
jobs_fp = '%s/%sjobs.txt' % (output_dir, job_prefix)
created_temp_paths.append(jobs_fp)
# generate the list of commands to be pushed out to nodes
commands, job_result_filepaths = \
get_job_commands(python_exe_fp,single_rarefaction_fp,job_prefix,\
input_path,output_dir,working_dir,min_seqs,max_seqs,step,num_reps,
lineages_included,command_prefix=' ',command_suffix=' ')
# Merge commands into jobs_to_start number of jobs
commands = merge_to_n_commands(commands,jobs_to_start)
# Set up poller apparatus if the user does not suppress polling
if not suppress_polling:
# Write the list of files which must exist for the jobs to be
# considered complete
expected_files_filepath = '%s/expected_out_files.txt' % working_dir
write_filepaths_to_file(job_result_filepaths,expected_files_filepath)
created_temp_paths.append(expected_files_filepath)
# Write the mapping file even though no merging is necessary
# (get_poller_command requires this, but a future version won't)
merge_map_filepath = '%s/merge_map.txt' % working_dir
open(merge_map_filepath,'w').close()
created_temp_paths.append(merge_map_filepath)
# Create the filepath listing the temporary files to be deleted,
# but don't write it yet
deletion_list_filepath = '%s/deletion_list.txt' % working_dir
created_temp_paths.append(deletion_list_filepath)
if not poll_directly:
# Generate the command to run the poller, and the list of temp files
# created by the poller
poller_command, poller_result_filepaths =\
get_poller_command(python_exe_fp,poller_fp,expected_files_filepath,\
merge_map_filepath,deletion_list_filepath,\
seconds_to_sleep=seconds_to_sleep)
# append the poller command to the list of job commands
commands.append(poller_command)
else:
poller_command, poller_result_filepaths =\
get_poller_command(python_exe_fp,poller_fp,expected_files_filepath,\
merge_map_filepath,deletion_list_filepath,\
seconds_to_sleep=seconds_to_sleep,\
command_prefix='',command_suffix='')
created_temp_paths += poller_result_filepaths
if not retain_temp_files:
# If the user wants temp files deleted, now write the list of
# temp files to be deleted
write_filepaths_to_file(created_temp_paths,deletion_list_filepath)
else:
# Otherwise just write an empty file
write_filepaths_to_file([],deletion_list_filepath)
# write the commands to the 'jobs files'
write_jobs_file(commands,job_prefix=job_prefix,jobs_fp=jobs_fp)
# submit the jobs file using cluster_jobs, if not suppressed by the
# user
if not opts.suppress_submit_jobs:
submit_jobs(path_to_cluster_jobs,jobs_fp,job_prefix)
if poll_directly:
try:
check_call(poller_command.split())
except CalledProcessError, e:
print '**Error occuring when calling the poller directly. '+\
'Jobs may have been submitted, but are not being polled.'
print str(e)
exit(-1)
def main():
option_parser, opts, args = parse_command_line_parameters(**script_info)
# create local copies of command-line options
python_exe_fp = opts.python_exe_fp
assign_taxonomy_fp = opts.assign_taxonomy_fp
confidence = opts.confidence
rdp_classifier_fp = opts.rdp_classifier_fp
id_to_taxonomy_fp = opts.id_to_taxonomy_fp
reference_seqs_fp = opts.reference_seqs_fp
cluster_jobs_fp = opts.cluster_jobs_fp
input_fasta_fp = opts.input_fasta_fp
jobs_to_start = opts.jobs_to_start
output_dir = opts.output_dir
poller_fp = opts.poller_fp
retain_temp_files = opts.retain_temp_files
suppress_polling = opts.suppress_polling
seconds_to_sleep = opts.seconds_to_sleep
poll_directly = opts.poll_directly
if not isfile(input_fasta_fp):
raise ValueError('This file does not exists: %s' % input_fasta_fp)
if id_to_taxonomy_fp or reference_seqs_fp:
if not id_to_taxonomy_fp or not isfile(id_to_taxonomy_fp):
raise ValueError('This file does not exits: %s' % id_to_taxonomy_fp)
if not reference_seqs_fp or not isfile(reference_seqs_fp):
raise ValueError('This file does not exits: %s' % reference_seqs_fp)
try:
makedirs(output_dir)
except OSError:
# output dir already exists
pass
created_temp_paths = []
# split the input filepath into directory and filename, base filename and
# extension
input_dir, input_fasta_fn = split(input_fasta_fp)
input_file_basename, input_fasta_ext = splitext(input_fasta_fn)
# set the job_prefix either based on what the user passed in,
# or a random string beginning with RDP
job_prefix = opts.job_prefix or get_random_job_prefix('RDP')
# A temporary output directory is created in output_dir named
# job_prefix. Output files are then moved from the temporary
# directory to the output directory when they are complete, allowing
# a poller to detect when runs complete by the presence of their
# output files.
working_dir = '%s/%s' % (output_dir,job_prefix)
try:
mkdir(working_dir)
created_temp_paths.append(working_dir)
except OSError:
# working dir already exists
pass
# compute the number of sequences that should be included in
# each file after splitting the input fasta file
num_seqs_per_file = compute_seqs_per_file(input_fasta_fp,jobs_to_start)
# split the fasta files and get the list of resulting files
tmp_fasta_fps =\
split_fasta(open(input_fasta_fp),num_seqs_per_file,job_prefix,output_dir)
created_temp_paths += tmp_fasta_fps
# build the filepath for the 'jobs script'
jobs_fp = '%s/%sjobs.txt' % (output_dir, job_prefix)
created_temp_paths.append(jobs_fp)
# generate the list of commands to be pushed out to nodes
commands, job_result_filepaths = \
get_commands(python_exe_fp,assign_taxonomy_fp,confidence,job_prefix,\
tmp_fasta_fps,rdp_classifier_fp,output_dir,working_dir,\
id_to_taxonomy_fp=id_to_taxonomy_fp,reference_seqs_fp=reference_seqs_fp)
created_temp_paths += job_result_filepaths
# Set up poller apparatus if the user does not suppress polling
if not suppress_polling:
# Write the list of files which must exist for the jobs to be
# considered complete
expected_files_filepath = '%s/expected_out_files.txt' % working_dir
write_filepaths_to_file(job_result_filepaths,expected_files_filepath)
created_temp_paths.append(expected_files_filepath)
# Write the mapping file which described how the output files from
# each job should be merged into the final output files
merge_map_filepath = '%s/merge_map.txt' % working_dir
write_merge_map_file_assign_taxonomy(job_result_filepaths,output_dir,\
merge_map_filepath,input_file_basename)
created_temp_paths.append(merge_map_filepath)
# Create the filepath listing the temporary files to be deleted,
# but don't write it yet
deletion_list_filepath = '%s/deletion_list.txt' % working_dir
created_temp_paths.append(deletion_list_filepath)
# Generate the command to run the poller, and the list of temp files
# created by the poller
if not poll_directly:
poller_command, poller_result_filepaths =\
get_poller_command(python_exe_fp,poller_fp,expected_files_filepath,\
merge_map_filepath,deletion_list_filepath,\
seconds_to_sleep=seconds_to_sleep)
created_temp_paths += poller_result_filepaths
# append the poller command to the list of job commands
commands.append(poller_command)
else:
poller_command, poller_result_filepaths =\
get_poller_command(python_exe_fp,poller_fp,\
expected_files_filepath,merge_map_filepath,\
deletion_list_filepath,seconds_to_sleep=seconds_to_sleep,\
command_prefix='',command_suffix='')
created_temp_paths += poller_result_filepaths
if not retain_temp_files:
# If the user wants temp files deleted, now write the list of
# temp files to be deleted
write_filepaths_to_file(created_temp_paths,deletion_list_filepath)
else:
# Otherwise just write an empty file
write_filepaths_to_file([],deletion_list_filepath)
# write the commands to the 'jobs files'
write_jobs_file(commands,job_prefix=job_prefix,jobs_fp=jobs_fp)
# submit the jobs file using cluster_jobs, if not suppressed by the
# user
if not opts.suppress_submit_jobs:
submit_jobs(cluster_jobs_fp,jobs_fp,job_prefix)
if poll_directly:
try:
check_call(poller_command.split())
except CalledProcessError, e:
print '**Error occuring when calling the poller directly. '+\
'Jobs may have been submitted, but are not being polled.'
print str(e)
exit(-1)
def main():
option_parser, opts, args = parse_command_line_parameters(**script_info)
# create local copies of command-line options
input_path = opts.input_path
output_dir = opts.output_path
metrics = opts.metrics
tree_fp = opts.tree_path
beta_diversity_fp = opts.beta_diversity_fp
python_exe_fp = opts.python_exe_fp
path_to_cluster_jobs = opts.cluster_jobs_fp
poller_fp = opts.poller_fp
retain_temp_files = opts.retain_temp_files
suppress_polling = opts.suppress_polling
seconds_to_sleep = opts.seconds_to_sleep
poll_directly = opts.poll_directly
jobs_to_start = opts.jobs_to_start
if isfile(input_path):
single_otu_table_mode = True
else:
single_otu_table_mode = False
input_fps = glob('%s/*' % input_path)
created_temp_paths = []
# split the input filepath into directory and filename, base filename and
# extension
# input_path, input_fn = split(input_path)
# input_file_basename, input_file_ext = splitext(input_fn)
# set the job_prefix either based on what the user passed in,
# or a random string beginning with BDIV
job_prefix = opts.job_prefix or get_random_job_prefix('BDIV')
# A temporary output directory is created in output_dir named
# job_prefix. Output files are then moved from the temporary
# directory to the output directory when they are complete, allowing
# a poller to detect when runs complete by the presence of their
# output files.
working_dir = '%s/%s' % (output_dir,job_prefix)
try:
makedirs(working_dir)
created_temp_paths.append(working_dir)
except OSError:
# working dir already exists
pass
# build the filepath for the 'jobs script'
jobs_fp = '%s/%sjobs.txt' % (output_dir, job_prefix)
created_temp_paths.append(jobs_fp)
# Get the list of commands to be run and the expected result files
if single_otu_table_mode:
# these will be the row dissim matrices
# temp for making, then move to output/i so the poller knows we're done
for i in range(jobs_to_start):
makedirs(working_dir + '/' + str(i))
created_temp_paths.append(working_dir + '/' + str(i))
makedirs(output_dir + '/' + str(i))
created_temp_paths.append(output_dir + '/' + str(i))
# to speed up this process, if not opts.full_tree: call setup here once
# and then use full_tree=True
# not implemented yet
commands, job_result_filepaths = \
get_job_commands_single_otu_table(python_exe_fp,beta_diversity_fp,
tree_fp,job_prefix,metrics,input_path,output_dir,working_dir,
jobs_to_start,command_prefix=' ',command_suffix=' ',
full_tree=opts.full_tree)
created_temp_paths += job_result_filepaths
else:
commands, job_result_filepaths = \
get_job_commands_multiple_otu_tables(python_exe_fp,beta_diversity_fp,
tree_fp,job_prefix,metrics,input_fps,output_dir,working_dir,
command_prefix=' ',command_suffix=' ', full_tree=opts.full_tree)
# Merge commands into jobs_to_start number of jobs
commands = merge_to_n_commands(commands,jobs_to_start)
# Set up poller apparatus if the user does not suppress polling
if not suppress_polling:
# Write the list of files which must exist for the jobs to be
# considered complete
expected_files_filepath = '%s/expected_out_files.txt' % working_dir
write_filepaths_to_file(job_result_filepaths,expected_files_filepath)
created_temp_paths.append(expected_files_filepath)
# Write the mapping file which described how the output files from
# each job should be merged into the final output files
merge_map_filepath = '%s/merge_map.txt' % working_dir
if single_otu_table_mode:
create_merge_map_file_single_otu_table(
input_path,output_dir,metrics,merge_map_filepath,
expected_files_filepath)
process_run_results_f =\
'qiime.parallel.beta_diversity.parallel_beta_diversity_process_run_results_f'
else:
open(merge_map_filepath,'w').close()
process_run_results_f = None
created_temp_paths.append(merge_map_filepath)
# Create the filepath listing the temporary files to be deleted,
# but don't write it yet
deletion_list_filepath = '%s/deletion_list.txt' % working_dir
created_temp_paths.append(deletion_list_filepath)
if not poll_directly:
# Generate the command to run the poller, and the list of temp files
# created by the poller
poller_command, poller_result_filepaths =\
get_poller_command(python_exe_fp,poller_fp,expected_files_filepath,
merge_map_filepath,deletion_list_filepath,
process_run_results_f=process_run_results_f,
seconds_to_sleep=seconds_to_sleep)
# append the poller command to the list of job commands
commands.append(poller_command)
else:
poller_command, poller_result_filepaths =\
get_poller_command(python_exe_fp,poller_fp,expected_files_filepath,
merge_map_filepath,deletion_list_filepath,
seconds_to_sleep=seconds_to_sleep,
process_run_results_f=process_run_results_f,
command_prefix='',command_suffix='')
created_temp_paths += poller_result_filepaths
if not retain_temp_files:
# If the user wants temp files deleted, now write the list of
# temp files to be deleted
write_filepaths_to_file(created_temp_paths,deletion_list_filepath)
else:
# Otherwise just write an empty file
write_filepaths_to_file([],deletion_list_filepath)
# write the commands to the 'jobs files'
write_jobs_file(commands,job_prefix=job_prefix,jobs_fp=jobs_fp)
# submit the jobs file using cluster_jobs, if not suppressed by the
# user
if not opts.suppress_submit_jobs:
submit_jobs(path_to_cluster_jobs,jobs_fp,job_prefix)
if poll_directly:
try:
check_call(poller_command.split())
except CalledProcessError, e:
print '**Error occuring when calling the poller directly. '+\
'Jobs may have been submitted, but are not being polled.'
print str(e)
exit(-1)
def __call__(self,
input_fp,
output_dir,
params,
job_prefix=None,
poll_directly=False,
suppress_submit_jobs=False):
""" """
## Generate a list of files and directories that will need to be cleaned up
self.files_to_remove = []
# split the input filepath into directory and filename, base filename and
# extension for use in naming other files
try:
input_dir, input_fn = split(input_fp)
input_file_basename, input_ext = splitext(input_fn)
except AttributeError:
## THIS IS AWFUL - SOME OF THE SCRIPTS PASS A LIST, SO THE
## PREVIOUS BLOCK WON'T WORK... WHAT DO WE WANT TO DO?
input_dir, input_fn = split(input_fp[0])
input_file_basename, input_ext = splitext(input_fn)
# Allow the user to override the default job_prefix (defined by the
# base classes)
if job_prefix is None:
job_prefix = get_random_job_prefix(self._job_prefix)
# A temporary output directory is created in output_dir named
# job_prefix. Output files are then moved from the temporary
# directory to the output directory when they are complete,
# allowing a poller to detect when runs complete by the presence
# of their output files.
working_dir = join(output_dir,job_prefix)
try:
makedirs(working_dir)
self.files_to_remove.append(working_dir)
except OSError:
# working dir already exists
pass
# Split the input file into the individual job input files. Add the
# individual job files to the files_to_remove list
input_fps, remove_input_on_completion = self._input_splitter(
input_fp,
self._jobs_to_start,
job_prefix,
working_dir)
if remove_input_on_completion:
self.files_to_remove += input_fps
# Perform any method-specific setup (e.g., formatting a BLAST database)
self._precommand_initiation(input_fp,output_dir,working_dir,params)
# Generate the list of commands to be pushed out to workers
# and the list of output files generated by each job.
commands, job_result_filepaths = self._get_job_commands(input_fps,
output_dir,
params,
job_prefix,
working_dir)
self.files_to_remove += \
self._identify_files_to_remove(job_result_filepaths,params)
# Generate the output clean-up files
merge_map_filepath, deletion_list_filepath, expected_files_filepath =\
self._initialize_output_cleanup_files(job_result_filepaths,
output_dir,
working_dir,
input_file_basename,
params)
# Set up poller apparatus if the user does not suppress polling
if not self._suppress_polling:
poller_command = self._initiate_polling(job_result_filepaths,
working_dir,
poll_directly,
merge_map_filepath,
deletion_list_filepath,
expected_files_filepath)
# If the poller should be run in the same way as the other commands
# (rather than by the current process), add it to the list of commands
if not poll_directly:
commands.append(poller_command)
# Build the filepath for the 'jobs script'. Add that file to the
# files_to_remove list.
jobs_fp = join(working_dir,job_prefix + 'jobs.txt')
self._write_jobs_file(commands,jobs_fp)
self.files_to_remove.append(jobs_fp)
# submit the jobs file using cluster_jobs, if not suppressed by the
# user
if not suppress_submit_jobs:
stdout, stderr, return_value = self._submit_jobs(
jobs_fp=jobs_fp, job_prefix=job_prefix)
# If the poller is going to be run by the current process,
# start polling
if poll_directly:
try:
check_call(poller_command.split())
except CalledProcessError, e:
print '**Error occuring when calling the poller directly. '+\
'Jobs may have been submitted, but are not being polled.'
print str(e)
print poller_command
exit(-1)
def main():
option_parser, opts, args = parse_command_line_parameters(**script_info)
# create local copies of command-line options
python_exe_fp = opts.python_exe_fp
pick_otus_fp = opts.pick_otus_fp
refseqs_fp = opts.refseqs_fp
cluster_jobs_fp = opts.cluster_jobs_fp
input_fasta_fp = opts.input_fasta_fp
jobs_to_start = opts.jobs_to_start
output_dir = opts.output_dir
poller_fp = opts.poller_fp
retain_temp_files = opts.retain_temp_files
suppress_polling = opts.suppress_polling
seconds_to_sleep = opts.seconds_to_sleep
similarity = opts.similarity
poll_directly = opts.poll_directly
uclust_stable_sort = not opts.suppress_uclust_stable_sort
save_uc_files = opts.save_uc_files
enable_rev_strand_match = opts.enable_rev_strand_match
optimal_uclust = opts.optimal_uclust
exact_uclust = opts.exact_uclust
max_accepts = opts.max_accepts
max_rejects = opts.max_rejects
stepwords = opts.stepwords
word_length = opts.word_length
created_temp_paths = []
# split the input filepath into directory and filename, base filename and
# extension
input_dir, input_fasta_fn = split(input_fasta_fp)
input_file_basename, input_fasta_ext = splitext(input_fasta_fn)
# set the job_prefix either based on what the user passed in,
# or a random string beginning with RDP
job_prefix = opts.job_prefix or get_random_job_prefix('POTU')
# A temporary output directory is created in output_dir named
# job_prefix. Output files are then moved from the temporary
# directory to the output directory when they are complete, allowing
# a poller to detect when runs complete by the presence of their
# output files.
working_dir = '%s/%s' % (output_dir,job_prefix)
try:
makedirs(working_dir)
created_temp_paths.append(working_dir)
except OSError:
# working dir already exists
pass
# compute the number of sequences that should be included in
# each file after splitting the input fasta file
num_seqs_per_file = compute_seqs_per_file(input_fasta_fp,jobs_to_start)
# split the fasta files and get the list of resulting files
tmp_fasta_fps =\
split_fasta(open(input_fasta_fp),num_seqs_per_file,\
job_prefix,working_dir=output_dir)
created_temp_paths += tmp_fasta_fps
# build the filepath for the 'jobs script'
jobs_fp = '%s/%sjobs.txt' % (output_dir, job_prefix)
created_temp_paths.append(jobs_fp)
# generate the list of commands to be pushed out to nodes and the list of
# output files generated by each job
commands, job_result_filepaths = \
get_job_commands(python_exe_fp,pick_otus_fp,tmp_fasta_fps,
output_dir,refseqs_fp,job_prefix,working_dir,similarity,
enable_rev_strand_match,optimal_uclust,exact_uclust,max_accepts,max_rejects,
stepwords, word_length, uclust_stable_sort, save_uc_files)
if save_uc_files:
# keep any .uc files that get created
created_temp_paths +=\
[fp for fp in job_result_filepaths if not fp.endswith('.uc')]
else:
created_temp_paths += [job_result_filepaths]
# Set up poller apparatus if the user does not suppress polling
if not suppress_polling:
# Write the list of files which must exist for the jobs to be
# considered complete
expected_files_filepath = '%s/expected_out_files.txt' % working_dir
write_filepaths_to_file(job_result_filepaths,expected_files_filepath)
created_temp_paths.append(expected_files_filepath)
# Write the mapping file which described how the output files from
# each job should be merged into the final output files
merge_map_filepath = '%s/merge_map.txt' % working_dir
process_run_results_f =\
'qiime.parallel.pick_otus_uclust_ref.parallel_uclust_ref_process_run_results_f'
write_merge_map_file_pick_otus(job_result_filepaths,output_dir,\
merge_map_filepath,input_file_basename,failures=True)
created_temp_paths.append(merge_map_filepath)
# Create the filepath listing the temporary files to be deleted,
# but don't write it yet
deletion_list_filepath = '%s/deletion_list.txt' % working_dir
created_temp_paths.append(deletion_list_filepath)
# Generate the command to run the poller, and the list of temp files
# created by the poller
if not poll_directly:
poller_command, poller_result_filepaths =\
get_poller_command(python_exe_fp,poller_fp,expected_files_filepath,\
merge_map_filepath,deletion_list_filepath,process_run_results_f,\
seconds_to_sleep=seconds_to_sleep)
created_temp_paths += poller_result_filepaths
# append the poller command to the list of job commands
commands.append(poller_command)
else:
poller_command, poller_result_filepaths =\
get_poller_command(python_exe_fp,poller_fp,expected_files_filepath,\
merge_map_filepath,deletion_list_filepath,process_run_results_f,\
seconds_to_sleep=seconds_to_sleep,command_prefix='',command_suffix='')
created_temp_paths += poller_result_filepaths
if not retain_temp_files:
# If the user wants temp files deleted, now write the list of
# temp files to be deleted
write_filepaths_to_file(created_temp_paths,deletion_list_filepath)
else:
# Otherwise just write an empty file
write_filepaths_to_file([],deletion_list_filepath)
# write the commands to the 'jobs files'
write_jobs_file(commands,job_prefix=job_prefix,jobs_fp=jobs_fp)
# submit the jobs file using cluster_jobs, if not suppressed by the
# user
if not opts.suppress_submit_jobs:
submit_jobs(cluster_jobs_fp,jobs_fp,job_prefix)
if poll_directly:
try:
check_call(poller_command.split())
except CalledProcessError, e:
print '**Error occuring when calling the poller directly. '+\
'Jobs may have been submitted, but are not being polled.'
print str(e)
exit(-1)
