def test_arrayjob(tmpdir):
"""Verify array jobs have multiple processes, separate log files for each task, and support parameter substitution..
"""
array_file_path = tmpdir.join("array_file")
array_file_path.write("World 1\nWorld 2\nWorld 3\n")
log_file_path = tmpdir.join("logfile.log")
runner = JobRunner("local")
runner.run_array("echo Hello {1} {2}", "JobName", str(log_file_path),
str(array_file_path))
assert (tmpdir.join("logfile.log-1").read() == "Hello World 1\n")
assert (tmpdir.join("logfile.log-2").read() == "Hello World 2\n")
assert (tmpdir.join("logfile.log-3").read() == "Hello World 3\n")
def test_quiet_array_run(tmpdir, capfd):
"""Verify NO tee output to stdout when not in quiet mode, and logfile captures all.
"""
array_file_path = tmpdir.join("array_file")
array_file_path.write("text\n")
log_file_path = tmpdir.join("logfile.log")
runner = JobRunner("local")
runner.run_array("(echo {1} to stdout; echo {1} to stderr 1>&2)",
"JobName",
str(log_file_path),
str(array_file_path),
quiet=True)
captured = capfd.readouterr()
assert (tmpdir.join("logfile.log-1").read() ==
'text to stdout\ntext to stderr\n')
assert (len(captured.out) == 0)
assert (len(captured.err) == 0)
def run(args):
"""Run all the steps of the snp pipeline in th correct order.
Parameters
----------
args : Namespace
referenceFile : str
Relative or absolute path to the reference fasta file
forceFlag : bool
Force processing even when result files already exist and are newer than inputs
mirror : str
Mode to create a mirror copy of the reference directory and all the sample directories.
Possible values: {soft, hard, copy}
configFile : str
Relative or absolute path to a configuration file for overriding defaults and defining
extra parameters for the tools and scripts within the pipeline.
jobQueueMgr : str
Job queue manager for remote parallel job execution in an HPC environment. Currently
"torque" and "grid" are supported. If not specified, the pipeline will execute locally.
workDir : str
Output directory for the result files.
samplesDir : str
Relative or absolute path to the parent directory of all the sample directories.
samplesFile : str
Relative or absolute path to a file listing all of the sample directories.
"""
global log_dir
global job_queue_mgr
start_time = time.time()
# Where are we running: grid, torque, or None (local)
job_queue_mgr = args.jobQueueMgr
# Erase any left-over error log environment variable from a previous run
os.environ.pop("errorOutputFile", None) # the 2nd arg avoids an exception when not in dict
# Handle output working directory. Create the directory if it does not exist.
# Any errors creating the work_dir will not be logged to the error log because
# the error log belongs in the work_dir.
work_dir = args.workDir
try:
utils.mkdir_p(work_dir)
except OSError as exc:
utils.fatal_error("Error: could not create the output directory %s" % work_dir)
if not utils.is_directory_writeable(work_dir):
utils.fatal_error("Error: output directory % is not writable." % work_dir)
# The error log is in the main workdir
error_output_file = os.path.join(work_dir, "error.log")
os.environ["errorOutputFile"] = error_output_file
# TODO: copy old error log to old logs directory, because otherwise it will be removed and lost forever
if os.path.isfile(error_output_file):
os.remove(error_output_file)
# Validate reference fasta file
reference_file_path = args.referenceFile
if not os.path.isfile(reference_file_path):
utils.fatal_error("Error: reference file %s does not exist." % reference_file_path)
if os.path.getsize(reference_file_path) == 0:
utils.fatal_error("Error: reference file %s is empty." % reference_file_path)
reference_file_name = os.path.basename(reference_file_path)
# Force rebuild flag is passed to all the subtask commands below
force_flag = " -f " if args.forceFlag else " "
# Create the logs directory with name like "logs-20170215.144253"
run_time_stamp = time.strftime('%Y%m%d.%H%M%S', time.localtime())
log_dir = os.path.join(work_dir, "logs-" + run_time_stamp)
try:
utils.mkdir_p(log_dir)
except OSError as exc:
utils.fatal_error("Error: could not create the logs directory %s" % log_dir)
if not utils.is_directory_writeable(work_dir):
utils.fatal_error("Error: logs directory % is not writable." % log_dir)
# Handle configuration file, use the specified file, or create a default file
if args.configFile:
config_file_path = args.configFile
if not os.path.isfile(config_file_path):
utils.fatal_error("Error: configuration file %s does not exist." % config_file_path)
if os.path.getsize(config_file_path) == 0:
utils.fatal_error("Error: configuration file %s is empty." % config_file_path)
shutil.copy2(config_file_path, log_dir) # copy2 tries to preserve timestamps
config_params = utils.read_properties(config_file_path, recognize_vars=True)
validate_properties(config_params)
else:
command.run("cfsan_snp_pipeline data configurationFile " + log_dir, outfile=sys.stdout)
config_file_path = os.path.join(log_dir, "snppipeline.conf")
config_params = utils.read_properties(config_file_path, recognize_vars=True)
# Validate the configured aligner choice
snp_pipeline_aligner = config_params.get("SnpPipeline_Aligner", "").lower() or "bowtie2"
if snp_pipeline_aligner not in ["bowtie2", "smalt"]:
utils.fatal_error("Config file error in SnpPipeline_Aligner parameter: only bowtie2 and smalt aligners are supported.")
os.environ["SnpPipeline_Aligner"] = snp_pipeline_aligner
# Stop the pipeline by default upon single sample errors if not configured either way
# The environment variable is used by called processes
stop_on_error = config_params.get("StopOnSampleError", "").lower() or "true"
os.environ["StopOnSampleError"] = stop_on_error
# Convert the stop_on_error flag to boolean for internal use in this function
stop_on_error = stop_on_error == "true"
# How many CPU cores can we use?
max_cpu_cores = config_params.get("MaxCpuCores", None)
if max_cpu_cores == "":
max_cpu_cores = None
if max_cpu_cores:
try:
max_cpu_cores = int(max_cpu_cores)
if max_cpu_cores < 1:
utils.fatal_error("Config file error in MaxCpuCores parameter: %s is less than one." % max_cpu_cores)
except ValueError:
utils.fatal_error("Config file error in MaxCpuCores parameter: %s is not a valid number." % max_cpu_cores)
if job_queue_mgr is None: # workstation
num_local_cpu_cores = psutil.cpu_count()
max_cpu_cores = min(num_local_cpu_cores, max_cpu_cores) if max_cpu_cores else num_local_cpu_cores
# How many CPU cores per process?
if job_queue_mgr is None: # workstation
cpu_cores_per_process = config_params.get("CpuCoresPerProcessOnWorkstation", None)
if cpu_cores_per_process:
try:
cpu_cores_per_process = int(cpu_cores_per_process)
if cpu_cores_per_process < 1:
utils.fatal_error("Config file error in CpuCoresPerProcessOnWorkstation parameter: %s is less than one." % cpu_cores_per_process)
except ValueError:
utils.fatal_error("Config file error in CpuCoresPerProcessOnWorkstation parameter: %s is not a valid number." % cpu_cores_per_process)
else:
cpu_cores_per_process = min(num_local_cpu_cores, max_cpu_cores)
else: # HPC
cpu_cores_per_process = config_params.get("CpuCoresPerProcessOnHPC", None)
if not cpu_cores_per_process:
utils.fatal_error("Config file error. CpuCoresPerProcessOnHPC parameter must be set to a value.")
else:
try:
cpu_cores_per_process = int(cpu_cores_per_process)
if cpu_cores_per_process < 1:
utils.fatal_error("Config file error in CpuCoresPerProcessOnHPC parameter: %s is less than one." % cpu_cores_per_process)
except ValueError:
utils.fatal_error("Config file error in CpuCoresPerProcessOnHPC parameter: %s is not a valid number." % cpu_cores_per_process)
# Put the configuration parameters into the process environment variables
os.environ["Bowtie2Build_ExtraParams"] = config_params.get("Bowtie2Build_ExtraParams", "")
os.environ["SmaltIndex_ExtraParams"] = config_params.get("SmaltIndex_ExtraParams", "")
os.environ["CreateSequenceDictionary_ExtraParams"] = config_params.get("CreateSequenceDictionary_ExtraParams", "")
os.environ["SamtoolsFaidx_ExtraParams"] = config_params.get("SamtoolsFaidx_ExtraParams", "")
os.environ["Bowtie2Align_ExtraParams"] = config_params.get("Bowtie2Align_ExtraParams", "")
os.environ["SmaltAlign_ExtraParams"] = config_params.get("SmaltAlign_ExtraParams", "")
os.environ["SamtoolsSamFilter_ExtraParams"] = config_params.get("SamtoolsSamFilter_ExtraParams", "")
os.environ["SamtoolsSort_ExtraParams"] = config_params.get("SamtoolsSort_ExtraParams", "")
os.environ["SamtoolsIndex_ExtraParams"] = config_params.get("SamtoolsIndex_ExtraParams", "")
os.environ["RemoveDuplicateReads"] = config_params.get("RemoveDuplicateReads", "").lower() or "true"
os.environ["PicardJvm_ExtraParams"] = config_params.get("PicardJvm_ExtraParams", "")
os.environ["PicardMarkDuplicates_ExtraParams"] = config_params.get("PicardMarkDuplicates_ExtraParams", "")
os.environ["EnableLocalRealignment"] = config_params.get("EnableLocalRealignment", "").lower() or "true"
os.environ["GatkJvm_ExtraParams"] = config_params.get("GatkJvm_ExtraParams", "")
os.environ["RealignerTargetCreator_ExtraParams"] = config_params.get("RealignerTargetCreator_ExtraParams", "")
os.environ["IndelRealigner_ExtraParams"] = config_params.get("IndelRealigner_ExtraParams", "")
os.environ["SamtoolsMpileup_ExtraParams"] = config_params.get("SamtoolsMpileup_ExtraParams", "")
os.environ["VarscanMpileup2snp_ExtraParams"] = config_params.get("VarscanMpileup2snp_ExtraParams", "")
os.environ["VarscanJvm_ExtraParams"] = config_params.get("VarscanJvm_ExtraParams", "")
os.environ["FilterRegions_ExtraParams"] = config_params.get("FilterRegions_ExtraParams", "")
os.environ["MergeSites_ExtraParams"] = config_params.get("MergeSites_ExtraParams", "")
os.environ["CallConsensus_ExtraParams"] = config_params.get("CallConsensus_ExtraParams", "")
os.environ["SnpMatrix_ExtraParams"] = config_params.get("SnpMatrix_ExtraParams", "")
os.environ["BcftoolsMerge_ExtraParams"] = config_params.get("BcftoolsMerge_ExtraParams", "")
os.environ["SnpReference_ExtraParams"] = config_params.get("SnpReference_ExtraParams", "")
os.environ["MergeVcfs_ExtraParams"] = config_params.get("MergeVcfs_ExtraParams", "")
os.environ["CollectMetrics_ExtraParams"] = config_params.get("CollectMetrics_ExtraParams", "")
os.environ["CombineMetrics_ExtraParams"] = config_params.get("CombineMetrics_ExtraParams", "")
# Verify the dependencies are available on the path
print("Checking dependencies...")
dependencies = ["cfsan_snp_pipeline", snp_pipeline_aligner, "java", "tabix", "bgzip", "bcftools"]
found_all_dependencies = True
for executable in dependencies:
if not utils.which(executable):
utils.report_error(executable + " is not on the path")
found_all_dependencies = False
if not utils.which("samtools"):
utils.report_error("samtools is not on the path")
found_all_dependencies = False
else:
version_str = utils.extract_version_str("SAMtools", "samtools 2>&1 > /dev/null")
samtools_version = version_str.split()[-1] # just the number
if samtools_version < "1.4":
utils.report_error("The installed %s is not supported. Version 1.4 or higher is required." % version_str)
found_all_dependencies = False
jar_file_path = utils.find_path_in_path_list("VarScan", "CLASSPATH")
if jar_file_path:
stdout = command.run("java -jar " + jar_file_path + " 2>&1")
if not jar_file_path or "error" in stdout.lower():
utils.report_error("CLASSPATH is not configured with the path to VarScan.jar")
found_all_dependencies = False
picard_required = os.environ["RemoveDuplicateReads"] == "true" or os.environ["EnableLocalRealignment"] == "true"
if picard_required:
jar_file_path = utils.find_path_in_path_list("picard", "CLASSPATH")
if not jar_file_path:
utils.report_error("CLASSPATH is not configured with the path to picard.jar")
found_all_dependencies = False
else:
stdout = command.run("java -jar " + jar_file_path + " 2>&1")
if stdout.lower().startswith("error"):
utils.report_error(stdout)
found_all_dependencies = False
gatk_required = os.environ["EnableLocalRealignment"] == "true"
if gatk_required:
jar_file_path = utils.find_path_in_path_list("GenomeAnalysisTK", "CLASSPATH")
if not jar_file_path:
utils.report_error("CLASSPATH is not configured with the path to GenomeAnalysisTK.jar")
found_all_dependencies = False
else:
stdout = command.run("java -jar " + jar_file_path + " --version 2>&1")
if stdout.lower().startswith("error"):
utils.report_error(stdout)
found_all_dependencies = False
else:
stdout = command.run("java -jar " + jar_file_path + " -T IndelRealigner --version 2>&1")
if "not a valid command" in stdout.lower() or "indelrealigner is no longer included" in stdout.lower():
utils.report_error("The installed GATK version does not support indel realignment. Try installing an older release prior to GATK v4.")
found_all_dependencies = False
elif "user error has occurred" in stdout.lower():
utils.report_error(stdout)
found_all_dependencies = False
if not found_all_dependencies:
utils.fatal_error("Check the SNP Pipeline installation instructions here: http://snp-pipeline.readthedocs.org/en/latest/installation.html")
else:
print("OK")
# Process the sample directory command line option
# TODO: detect broken fastq symlinks
if args.samplesDir:
samples_parent_dir = args.samplesDir.rstrip('/') # strip trailing slash
if not utils.verify_non_empty_directory("Samples directory", samples_parent_dir):
sys.exit(1)
# verify at least one of the subdirectories contains fastq files.
dir_sizes = get_sorted_sample_dirs_fastq_sizes(samples_parent_dir)
dir_sizes = [(size, path) for size, path in dir_sizes if size > 0]
if len(dir_sizes) == 0:
utils.fatal_error("Samples directory %s does not contain subdirectories with fastq files." % samples_parent_dir)
sample_dirs_file = os.path.join(work_dir, "sampleDirectories.txt")
persist_sorted_sample_dirs_file(samples_parent_dir, sample_dirs_file)
# Process the file of sample directories command line option
# TODO: detect broken fastq symlinks
if args.samplesFile:
sample_dirs_file = args.samplesFile
if not os.path.isfile(sample_dirs_file):
utils.fatal_error("Error: the file of samples directories, %s, does not exist." % sample_dirs_file)
if os.path.getsize(sample_dirs_file) == 0:
utils.fatal_error("Error: the file of samples directories, %s, is empty." % sample_dirs_file)
rewrite_cleansed_file_of_sample_dirs(sample_dirs_file, os.path.join(work_dir, "sampleDirectories.txt"))
sample_dirs_file = os.path.join(work_dir, "sampleDirectories.txt")
validate_file_of_sample_dirs(sample_dirs_file)
with open(sample_dirs_file) as f:
sample_dirs_list = f.read().splitlines()
sample_count = len(sample_dirs_list)
# --------------------------------------------------------
if job_queue_mgr is None:
progress("Step 1 - Prep work")
else:
print("Step 1 - Prep work")
# --------------------------------------------------------
# Mirror the input reference and samples if requested
# TODO: make this a pure python solution
if args.mirror:
if args.mirror == "soft":
# soft link, subsequent freshness checks use the timestamp of original file, not the soft link
mirror_flag = " -s "
elif args.mirror == "hard":
# hard link, automatically preserves attributes of the original file
mirror_flag = " -l "
else:
# regular copy, -p explicitly preserves attributes of the original file
mirror_flag = " -p "
# flush stdout to keep the unbuffered stderr in chronological order with stdout
sys.stdout.flush()
# Mirror/link the reference
work_reference_dir = os.path.join(work_dir, "reference")
utils.mkdir_p(work_reference_dir)
src_reference_file = os.path.abspath(reference_file_path)
cmd = "cp -v -u -f" + mirror_flag + src_reference_file + ' ' + work_reference_dir
subprocess.check_call(cmd, shell=True)
# since we mirrored the reference, we need to update our reference location
reference_file_path = os.path.join(work_reference_dir, reference_file_name)
# Mirror/link the samples
work_samples_parent_dir = os.path.join(work_dir, "samples")
for directory in sample_dirs_list:
basedir = os.path.basename(directory)
work_sample_dir = os.path.join(work_samples_parent_dir, basedir)
utils.mkdir_p(work_sample_dir)
src_sample_dir = os.path.abspath(directory)
# copy without stderr message and without exit error code because the fastq or fq files might not exist
cmd = "cp -r -v -u -f" + mirror_flag + src_sample_dir + "/*.fastq* " + work_sample_dir + " 2> /dev/null || true"
subprocess.check_call(cmd, shell=True)
cmd = "cp -r -v -u -f" + mirror_flag + src_sample_dir + "/*.fq* " + work_sample_dir + " 2> /dev/null || true"
subprocess.check_call(cmd, shell=True)
# since we mirrored the samples, we need to update our sorted list of samples
sample_dirs_file = os.path.join(work_dir, "sampleDirectories.txt")
persist_sorted_sample_dirs_file(work_samples_parent_dir, sample_dirs_file)
# refresh the list of sample dirs -- now in sorted order
with open(sample_dirs_file) as f:
sample_dirs_list = f.read().splitlines()
# get the *.fastq or *.fq files in each sample directory, possibly compresessed, on one line per sample, ready to feed to bowtie
sample_full_path_names_file = os.path.join(work_dir, "sampleFullPathNames.txt")
with open(sample_full_path_names_file, 'w') as f:
for directory in sample_dirs_list:
file_list = fastq.list_fastq_files(directory)
print(' '.join(file_list), file=f)
# Initialize the job runner
if job_queue_mgr is None:
runner = JobRunner("local", exception_handler=handle_exception, verbose=args.verbose >= 4)
elif job_queue_mgr == "grid":
strip_job_array_suffix = config_params.get("GridEngine_StripJobArraySuffix", "true").lower()
qsub_extra_params = config_params.get("GridEngine_QsubExtraParams")
runner = JobRunner(job_queue_mgr, strip_job_array_suffix == "true", qsub_extra_params=qsub_extra_params, verbose=args.verbose >= 4)
else:
strip_job_array_suffix = config_params.get("Torque_StripJobArraySuffix", "false").lower()
qsub_extra_params = config_params.get("Torque_QsubExtraParams")
runner = JobRunner(job_queue_mgr, strip_job_array_suffix == "true", qsub_extra_params=qsub_extra_params, verbose=args.verbose >= 4)
progress("Step 2 - Index the reference")
log_file = os.path.join(log_dir, "indexRef.log")
command_line = "cfsan_snp_pipeline index_ref" + force_flag + reference_file_path
job_id_index_ref = runner.run(command_line, "indexRef", log_file)
progress("Step 3 - Map the sample reads to the reference")
# Parse the user-specified aligner parameters to find the number of CPU cores requested, for example, "-p 16" or "-n 16"
# Set the default number of CPU cores if the user did not configure a value.
if snp_pipeline_aligner == "smalt":
extra_params_env_var = "SmaltAlign_ExtraParams"
threads_option = "-n"
else:
extra_params_env_var = "Bowtie2Align_ExtraParams"
threads_option = "-p"
aligner_max_processes, aligner_threads_per_process = utils.configure_process_threads(extra_params_env_var, threads_option, cpu_cores_per_process, max_cpu_cores)
samfilter_max_processes, samfilter_threads_per_process = utils.configure_process_threads("SamtoolsSamFilter_ExtraParams", ["-@", "--threads"], cpu_cores_per_process, max_cpu_cores)
samsort_max_processes, samsort_threads_per_process = utils.configure_process_threads("SamtoolsSort_ExtraParams", ["-@", "--threads"], cpu_cores_per_process, max_cpu_cores)
samindex_max_processes, samindex_threads_per_process = utils.configure_process_threads("SamtoolsIndex_ExtraParams", ["-@"], cpu_cores_per_process, max_cpu_cores)
realigner_max_processes, realigner_threads_per_process = utils.configure_process_threads("RealignerTargetCreator_ExtraParams", ["-nt", "--num_threads"], cpu_cores_per_process, max_cpu_cores)
# There are multiple processes within map_reads, each with multiple threads.
# The CPU allocation must be enough for the process needing the largest number of threads.
max_processes_list = [aligner_max_processes, samfilter_max_processes, samsort_max_processes, samindex_max_processes, realigner_max_processes]
if all([i is None for i in max_processes_list]):
max_processes = None
else:
max_processes = min([i for i in max_processes_list if i is not None])
threads_per_process = max(aligner_threads_per_process, samfilter_threads_per_process, samsort_threads_per_process, samindex_threads_per_process, realigner_threads_per_process)
parallel_environment = config_params.get("GridEngine_PEname", None)
log_file = os.path.join(log_dir, "mapReads.log")
command_line = "cfsan_snp_pipeline map_reads --threads " + str(threads_per_process) + force_flag + reference_file_path + " {1} {2}"
job_id_map_reads = runner.run_array(command_line, "mapReads", log_file, sample_full_path_names_file, max_processes=max_processes, wait_for=[job_id_index_ref], threads=threads_per_process, parallel_environment=parallel_environment)
progress("Step 4 - Find sites with SNPs in each sample")
if job_queue_mgr in ["grid", "torque"]:
time.sleep(1.0 + float(sample_count) / 150) # workaround torque bug when submitting two large consecutive array jobs, potential bug for grid
log_file = os.path.join(log_dir, "callSites.log")
command_line = "cfsan_snp_pipeline call_sites" + force_flag + reference_file_path + " {1}"
job_id_call_sites = runner.run_array(command_line, "callSites", log_file, sample_dirs_file, max_processes=max_cpu_cores, wait_for_array=[job_id_map_reads], slot_dependency=True)
progress("Step 5 - Filter abnormal SNP regions")
log_file = os.path.join(log_dir, "filterRegions.log")
extra_params = os.environ.get("FilterRegions_ExtraParams", "")
command_line = "cfsan_snp_pipeline filter_regions" + force_flag + "-n var.flt.vcf " + sample_dirs_file + ' ' + reference_file_path + ' ' + extra_params
job_id_filter_regions = runner.run(command_line, "filterRegions", log_file, wait_for_array=[job_id_call_sites])
# Starting from here, there are 2 threads:
# Thread X.1: the thread processing the original VCF files and corresponding downstream results
# Thread X.2: the thread processing the preserved VCF files and corresponding downstream results
progress("Step 6.1 - Merge the SNP sites across all samples into the SNP list file")
# The mergeSites process creates the filtered list of sample directories. It is the list of samples not having excessive snps.
# When running on a workstation, the file exists at this point during the script execution, but on grid or torque, it has not yet been created. However,
# we know the path to the file regardless of whether it exists yet.
filtered_sample_dirs_file = sample_dirs_file + ".OrigVCF.filtered"
# touch $filtered_sample_dirs_file # TODO: why was this touch here in the old run_snp_pipeline.sh script?
log_file = os.path.join(log_dir, "mergeSites.log")
output_file = os.path.join(work_dir, "snplist.txt")
extra_params = os.environ.get("MergeSites_ExtraParams", "")
command_line = "cfsan_snp_pipeline merge_sites" + force_flag + "-n var.flt.vcf -o " + output_file + ' ' + extra_params + ' ' + sample_dirs_file + ' ' + filtered_sample_dirs_file
job_id_merge_sites = runner.run(command_line, "mergeSites", log_file, wait_for=[job_id_filter_regions])
progress("Step 6.2 - Merge the SNP sites across all samples into the SNP list file")
# Create another copy of sample directories file, for the thread processing preserved snp files.
filtered_sample_dirs_file2 = sample_dirs_file + ".PresVCF.filtered"
# touch $filtered_sample_dirs_file2 # TODO: why was this touch here in the old run_snp_pipeline.sh script?
log_file = os.path.join(log_dir, "mergeSites_preserved.log")
output_file = os.path.join(work_dir, "snplist_preserved.txt")
extra_params = os.environ.get("MergeSites_ExtraParams", "")
command_line = "cfsan_snp_pipeline merge_sites" + force_flag + "-n var.flt_preserved.vcf -o " + output_file + ' ' + extra_params + ' ' + sample_dirs_file + ' ' + filtered_sample_dirs_file2
job_id_merge_sites2 = runner.run(command_line, "mergeSites_preserved", log_file, wait_for=[job_id_filter_regions])
progress("Step 7.1 - Call the consensus SNPs for each sample")
log_file = os.path.join(log_dir, "callConsensus.log")
list_file = os.path.join(work_dir, "snplist.txt")
output_file = "{1}/consensus.fasta"
extra_params = os.environ.get("CallConsensus_ExtraParams", "")
command_line = "cfsan_snp_pipeline call_consensus" + force_flag + "-l " + list_file + " -o " + output_file + " --vcfRefName " + reference_file_name + ' ' + extra_params + " --vcfFileName consensus.vcf {1}/reads.all.pileup"
job_id_call_consensus = runner.run_array(command_line, "callConsensus", log_file, sample_dirs_file, max_processes=max_cpu_cores, wait_for=[job_id_merge_sites])
progress("Step 7.2 - Call the consensus SNPs for each sample")
log_file = os.path.join(log_dir, "callConsensus_preserved.log")
list_file = os.path.join(work_dir, "snplist_preserved.txt")
output_file = "{1}/consensus_preserved.fasta"
extra_params = os.environ.get("CallConsensus_ExtraParams", "")
command_line = "cfsan_snp_pipeline call_consensus" + force_flag + "-l " + list_file + " -o " + output_file + " -e {1}/var.flt_removed.vcf --vcfRefName " + reference_file_name + ' ' + extra_params + " --vcfFileName consensus_preserved.vcf {1}/reads.all.pileup"
job_id_call_consensus2 = runner.run_array(command_line, "callConsensus_preserved", log_file, sample_dirs_file, max_processes=max_cpu_cores, wait_for=[job_id_merge_sites2])
progress("Step 8.1 - Create the SNP matrix")
log_file = os.path.join(log_dir, "snpMatrix.log")
output_file = os.path.join(work_dir, "snpma.fasta")
extra_params = os.environ.get("SnpMatrix_ExtraParams", "")
command_line = "cfsan_snp_pipeline snp_matrix" + force_flag + "-c consensus.fasta -o " + output_file + ' ' + extra_params + ' ' + filtered_sample_dirs_file
job_id_snp_matrix = runner.run(command_line, "snpMatrix", log_file, wait_for_array=[job_id_call_consensus])
progress("Step 8.2 - Create the SNP matrix")
log_file = os.path.join(log_dir, "snpMatrix_preserved.log")
output_file = os.path.join(work_dir, "snpma_preserved.fasta")
extra_params = os.environ.get("SnpMatrix_ExtraParams", "")
command_line = "cfsan_snp_pipeline snp_matrix" + force_flag + "-c consensus_preserved.fasta -o " + output_file + ' ' + extra_params + ' ' + filtered_sample_dirs_file2
job_id_snp_matrix2 = runner.run(command_line, "snpMatrix_preserved", log_file, wait_for_array=[job_id_call_consensus2])
progress("Step 9.1 - Create the reference sequence at SNP sites")
log_file = os.path.join(log_dir, "snpReference.log")
list_file = os.path.join(work_dir, "snplist.txt")
output_file = os.path.join(work_dir, "referenceSNP.fasta")
extra_params = os.environ.get("SnpReference_ExtraParams", "")
command_line = "cfsan_snp_pipeline snp_reference" + force_flag + "-l " + list_file + " -o " + output_file + ' ' + extra_params + ' ' + reference_file_path
job_id_snp_reference = runner.run(command_line, "snpReference", log_file, wait_for_array=[job_id_call_consensus])
progress("Step 9.2 - Create the reference sequence at SNP sites")
log_file = os.path.join(log_dir, "snpReference_preserved.log")
list_file = os.path.join(work_dir, "snplist_preserved.txt")
output_file = os.path.join(work_dir, "referenceSNP_preserved.fasta")
extra_params = os.environ.get("SnpReference_ExtraParams", "")
command_line = "cfsan_snp_pipeline snp_reference" + force_flag + "-l " + list_file + " -o " + output_file + ' ' + extra_params + ' ' + reference_file_path
job_id_snp_reference2 = runner.run(command_line, "snpReference_preserved", log_file, wait_for_array=[job_id_call_consensus2])
progress("Step 10.1 - Merge sample VCFs to create the multi-VCF file")
if "--vcfFileName" in os.environ.get("CallConsensus_ExtraParams", ""):
log_file = os.path.join(log_dir, "mergeVcfs.log")
output_file = os.path.join(work_dir, "snpma.vcf")
extra_params = os.environ.get("MergeVcfs_ExtraParams", "")
command_line = "cfsan_snp_pipeline merge_vcfs" + force_flag + "-o " + output_file + ' ' + extra_params + ' ' + filtered_sample_dirs_file
job_id_merge_vcfs = runner.run(command_line, "mergeVcfs", log_file, wait_for_array=[job_id_call_consensus])
else:
print("Skipped per CallConsensus_ExtraParams configuration")
progress("Step 10.2 - Merge sample VCFs to create the multi-VCF file")
if "--vcfFileName" in os.environ.get("CallConsensus_ExtraParams", ""):
log_file = os.path.join(log_dir, "mergeVcfs_preserved.log")
output_file = os.path.join(work_dir, "snpma_preserved.vcf")
extra_params = os.environ.get("MergeVcfs_ExtraParams", "")
command_line = "cfsan_snp_pipeline merge_vcfs" + force_flag + "-n consensus_preserved.vcf -o " + output_file + ' ' + extra_params + ' ' + filtered_sample_dirs_file2
job_id_merge_vcfs2 = runner.run(command_line, "mergeVcfs_preserved", log_file, wait_for_array=[job_id_call_consensus2])
else:
print("Skipped per CallConsensus_ExtraParams configuration")
progress("Step 11.1 - Calculate SNP distance matrix")
log_file = os.path.join(log_dir, "distance.log")
input_file = os.path.join(work_dir, "snpma.fasta")
pair_output_file = os.path.join(work_dir, "snp_distance_pairwise.tsv")
matrix_output_file = os.path.join(work_dir, "snp_distance_matrix.tsv")
command_line = "cfsan_snp_pipeline distance" + force_flag + "-p " + pair_output_file + " -m " + matrix_output_file + ' ' + input_file
job_id_distance = runner.run(command_line, "distance", log_file, wait_for=[job_id_snp_matrix])
progress("Step 11.2 - Calculate SNP distance matrix")
log_file = os.path.join(log_dir, "distance_preserved.log")
input_file = os.path.join(work_dir, "snpma_preserved.fasta")
pair_output_file = os.path.join(work_dir, "snp_distance_pairwise_preserved.tsv")
matrix_output_file = os.path.join(work_dir, "snp_distance_matrix_preserved.tsv")
command_line = "cfsan_snp_pipeline distance" + force_flag + "-p " + pair_output_file + " -m " + matrix_output_file + ' ' + input_file
job_id_distance2 = runner.run(command_line, "distance_preserved", log_file, wait_for=[job_id_snp_matrix2])
progress("Step 12 - Collect metrics for each sample")
log_file = os.path.join(log_dir, "collectMetrics.log")
output_file = "{1}/metrics"
extra_params = os.environ.get("CollectMetrics_ExtraParams", "")
command_line = "cfsan_snp_pipeline collect_metrics" + force_flag + "-o " + output_file + ' ' + extra_params + " {1} " + reference_file_path
job_id_collect_metrics = runner.run_array(command_line, "collectMetrics", log_file, sample_dirs_file, max_processes=max_cpu_cores, wait_for_array=[job_id_call_consensus, job_id_call_consensus2], slot_dependency=True)
progress("Step 13 - Combine the metrics across all samples into the metrics table")
log_file = os.path.join(log_dir, "combineMetrics.log")
output_file = os.path.join(work_dir, "metrics.tsv")
extra_params = os.environ.get("CombineMetrics_ExtraParams", "")
command_line = "cfsan_snp_pipeline combine_metrics" + force_flag + "-n metrics -o " + output_file + ' ' + extra_params + ' ' + sample_dirs_file
combine_metrics_job_id = runner.run(command_line, "combineMetrics", log_file, wait_for_array=[job_id_collect_metrics])
# Step 14 - Notify user of any non-fatal errors accumulated during processing
if os.path.isfile(error_output_file) and os.path.getsize(error_output_file) > 0 and not stop_on_error:
print("\nThere were errors processing some samples.\nSee the log file %s for a summary of errors." % error_output_file, file=sys.stderr)
# Exit here to prevent showing the "cfsan_snp_pipeline run finished" message. The jobs are queued, not finished yet.
if job_queue_mgr is not None: # HPC
sys.exit(0)
else:
end_time = time.time()
elapsed_time = end_time - start_time
print("Elapsed time =", elapsed_time)
def run(args):
"""Run all the steps of the snp pipeline in th correct order.
Parameters
----------
args : Namespace
referenceFile : str
Relative or absolute path to the reference fasta file
forceFlag : bool
Force processing even when result files already exist and are newer than inputs
mirror : str
Mode to create a mirror copy of the reference directory and all the sample directories.
Possible values: {soft, hard, copy}
configFile : str
Relative or absolute path to a configuration file for overriding defaults and defining
extra parameters for the tools and scripts within the pipeline.
jobQueueMgr : str
Job queue manager for remote parallel job execution in an HPC environment. Currently
"torque" and "grid" are supported. If not specified, the pipeline will execute locally.
workDir : str
Output directory for the result files.
samplesDir : str
Relative or absolute path to the parent directory of all the sample directories.
samplesFile : str
Relative or absolute path to a file listing all of the sample directories.
purge : bool
Purge the intermediate output files when the pipeline completes successfully.
"""
global log_dir
global job_queue_mgr
start_time = time.time()
# Where are we running: grid, torque, or None (local)
job_queue_mgr = args.jobQueueMgr
# Erase any left-over error log environment variable from a previous run
os.environ.pop("errorOutputFile",
None) # the 2nd arg avoids an exception when not in dict
# Handle output working directory. Create the directory if it does not exist.
# Any errors creating the work_dir will not be logged to the error log because
# the error log belongs in the work_dir.
work_dir = args.workDir
try:
utils.mkdir_p(work_dir)
except OSError as exc:
utils.fatal_error("Error: could not create the output directory %s" %
work_dir)
if not utils.is_directory_writeable(work_dir):
utils.fatal_error("Error: output directory % is not writable." %
work_dir)
# The error log is in the main workdir
error_output_file = os.path.join(work_dir, "error.log")
os.environ["errorOutputFile"] = error_output_file
# TODO: copy old error log to old logs directory, because otherwise it will be removed and lost forever
if os.path.isfile(error_output_file):
os.remove(error_output_file)
# Validate reference fasta file
reference_file_path = args.referenceFile
if not os.path.isfile(reference_file_path):
utils.fatal_error("Error: reference file %s does not exist." %
reference_file_path)
if os.path.getsize(reference_file_path) == 0:
utils.fatal_error("Error: reference file %s is empty." %
reference_file_path)
reference_file_name = os.path.basename(reference_file_path)
# Force rebuild flag is passed to all the subtask commands below
force_flag = " -f " if args.forceFlag else " "
# Create the logs directory with name like "logs-20170215.144253"
run_time_stamp = time.strftime('%Y%m%d.%H%M%S', time.localtime())
log_dir = os.path.join(work_dir, "logs-" + run_time_stamp)
try:
utils.mkdir_p(log_dir)
except OSError as exc:
utils.fatal_error("Error: could not create the logs directory %s" %
log_dir)
if not utils.is_directory_writeable(work_dir):
utils.fatal_error("Error: logs directory % is not writable." % log_dir)
# Handle configuration file, use the specified file, or create a default file
if args.configFile:
config_file_path = args.configFile
if not os.path.isfile(config_file_path):
utils.fatal_error("Error: configuration file %s does not exist." %
config_file_path)
if os.path.getsize(config_file_path) == 0:
utils.fatal_error("Error: configuration file %s is empty." %
config_file_path)
shutil.copy2(config_file_path,
log_dir) # copy2 tries to preserve timestamps
config_params = utils.read_properties(config_file_path,
recognize_vars=True)
validate_properties(config_params)
else:
command.run("cfsan_snp_pipeline data configurationFile " + log_dir,
outfile=sys.stdout)
config_file_path = os.path.join(log_dir, "snppipeline.conf")
config_params = utils.read_properties(config_file_path,
recognize_vars=True)
# Validate the configured aligner choice
snp_pipeline_aligner = config_params.get("SnpPipeline_Aligner",
"").lower() or "bowtie2"
if snp_pipeline_aligner not in ["bowtie2", "smalt"]:
utils.fatal_error(
"Config file error in SnpPipeline_Aligner parameter: only bowtie2 and smalt aligners are supported."
)
os.environ["SnpPipeline_Aligner"] = snp_pipeline_aligner
# Stop the pipeline by default upon single sample errors if not configured either way
# The environment variable is used by called processes
stop_on_error = config_params.get("StopOnSampleError",
"").lower() or "true"
os.environ["StopOnSampleError"] = stop_on_error
# Convert the stop_on_error flag to boolean for internal use in this function
stop_on_error = stop_on_error == "true"
# How many CPU cores can we use?
max_cpu_cores = config_params.get("MaxCpuCores", None)
if max_cpu_cores == "":
max_cpu_cores = None
if max_cpu_cores:
try:
max_cpu_cores = int(max_cpu_cores)
if max_cpu_cores < 1:
utils.fatal_error(
"Config file error in MaxCpuCores parameter: %s is less than one."
% max_cpu_cores)
except ValueError:
utils.fatal_error(
"Config file error in MaxCpuCores parameter: %s is not a valid number."
% max_cpu_cores)
if job_queue_mgr is None: # workstation
num_local_cpu_cores = psutil.cpu_count()
max_cpu_cores = min(
num_local_cpu_cores,
max_cpu_cores) if max_cpu_cores else num_local_cpu_cores
# How many CPU cores per process?
if job_queue_mgr is None: # workstation
cpu_cores_per_process = config_params.get(
"CpuCoresPerProcessOnWorkstation", None)
if cpu_cores_per_process:
try:
cpu_cores_per_process = int(cpu_cores_per_process)
if cpu_cores_per_process < 1:
utils.fatal_error(
"Config file error in CpuCoresPerProcessOnWorkstation parameter: %s is less than one."
% cpu_cores_per_process)
except ValueError:
utils.fatal_error(
"Config file error in CpuCoresPerProcessOnWorkstation parameter: %s is not a valid number."
% cpu_cores_per_process)
else:
cpu_cores_per_process = min(num_local_cpu_cores, max_cpu_cores)
else: # HPC
cpu_cores_per_process = config_params.get("CpuCoresPerProcessOnHPC",
None)
if not cpu_cores_per_process:
utils.fatal_error(
"Config file error. CpuCoresPerProcessOnHPC parameter must be set to a value."
)
else:
try:
cpu_cores_per_process = int(cpu_cores_per_process)
if cpu_cores_per_process < 1:
utils.fatal_error(
"Config file error in CpuCoresPerProcessOnHPC parameter: %s is less than one."
% cpu_cores_per_process)
except ValueError:
utils.fatal_error(
"Config file error in CpuCoresPerProcessOnHPC parameter: %s is not a valid number."
% cpu_cores_per_process)
# Put the configuration parameters into the process environment variables
os.environ["Bowtie2Build_ExtraParams"] = config_params.get(
"Bowtie2Build_ExtraParams", "")
os.environ["SmaltIndex_ExtraParams"] = config_params.get(
"SmaltIndex_ExtraParams", "")
os.environ["CreateSequenceDictionary_ExtraParams"] = config_params.get(
"CreateSequenceDictionary_ExtraParams", "")
os.environ["SamtoolsFaidx_ExtraParams"] = config_params.get(
"SamtoolsFaidx_ExtraParams", "")
os.environ["Bowtie2Align_ExtraParams"] = config_params.get(
"Bowtie2Align_ExtraParams", "")
os.environ["SmaltAlign_ExtraParams"] = config_params.get(
"SmaltAlign_ExtraParams", "")
os.environ["SamtoolsSamFilter_ExtraParams"] = config_params.get(
"SamtoolsSamFilter_ExtraParams", "")
os.environ["SamtoolsSort_ExtraParams"] = config_params.get(
"SamtoolsSort_ExtraParams", "")
os.environ["SamtoolsIndex_ExtraParams"] = config_params.get(
"SamtoolsIndex_ExtraParams", "")
os.environ["RemoveDuplicateReads"] = config_params.get(
"RemoveDuplicateReads", "").lower() or "true"
os.environ["PicardJvm_ExtraParams"] = config_params.get(
"PicardJvm_ExtraParams", "")
os.environ["PicardMarkDuplicates_ExtraParams"] = config_params.get(
"PicardMarkDuplicates_ExtraParams", "")
os.environ["EnableLocalRealignment"] = config_params.get(
"EnableLocalRealignment", "").lower() or "true"
os.environ["GatkJvm_ExtraParams"] = config_params.get(
"GatkJvm_ExtraParams", "")
os.environ["RealignerTargetCreator_ExtraParams"] = config_params.get(
"RealignerTargetCreator_ExtraParams", "")
os.environ["IndelRealigner_ExtraParams"] = config_params.get(
"IndelRealigner_ExtraParams", "")
os.environ["SamtoolsMpileup_ExtraParams"] = config_params.get(
"SamtoolsMpileup_ExtraParams", "")
os.environ["VarscanMpileup2snp_ExtraParams"] = config_params.get(
"VarscanMpileup2snp_ExtraParams", "")
os.environ["VarscanJvm_ExtraParams"] = config_params.get(
"VarscanJvm_ExtraParams", "")
os.environ["FilterRegions_ExtraParams"] = config_params.get(
"FilterRegions_ExtraParams", "")
os.environ["MergeSites_ExtraParams"] = config_params.get(
"MergeSites_ExtraParams", "")
os.environ["CallConsensus_ExtraParams"] = config_params.get(
"CallConsensus_ExtraParams", "")
os.environ["SnpMatrix_ExtraParams"] = config_params.get(
"SnpMatrix_ExtraParams", "")
os.environ["BcftoolsMerge_ExtraParams"] = config_params.get(
"BcftoolsMerge_ExtraParams", "")
os.environ["SnpReference_ExtraParams"] = config_params.get(
"SnpReference_ExtraParams", "")
os.environ["MergeVcfs_ExtraParams"] = config_params.get(
"MergeVcfs_ExtraParams", "")
os.environ["CollectMetrics_ExtraParams"] = config_params.get(
"CollectMetrics_ExtraParams", "")
os.environ["CombineMetrics_ExtraParams"] = config_params.get(
"CombineMetrics_ExtraParams", "")
# Verify the dependencies are available on the path
print("Checking dependencies...")
dependencies = [
"cfsan_snp_pipeline", snp_pipeline_aligner, "java", "tabix", "bgzip",
"bcftools"
]
found_all_dependencies = True
for executable in dependencies:
if not utils.which(executable):
utils.report_error(executable + " is not on the path")
found_all_dependencies = False
if not utils.which("samtools"):
utils.report_error("samtools is not on the path")
found_all_dependencies = False
else:
version_str = utils.extract_version_str("SAMtools",
"samtools 2>&1 > /dev/null")
samtools_version = version_str.split()[-1] # just the number
if samtools_version < "1.4":
utils.report_error(
"The installed %s is not supported. Version 1.4 or higher is required."
% version_str)
found_all_dependencies = False
jar_file_path = utils.find_path_in_path_list("VarScan", "CLASSPATH")
if jar_file_path:
stdout = command.run("java -jar " + jar_file_path + " 2>&1")
if not jar_file_path or "error" in stdout.lower():
utils.report_error(
"CLASSPATH is not configured with the path to VarScan.jar")
found_all_dependencies = False
picard_required = os.environ[
"RemoveDuplicateReads"] == "true" or os.environ[
"EnableLocalRealignment"] == "true"
if picard_required:
jar_file_path = utils.find_path_in_path_list("picard", "CLASSPATH")
if not jar_file_path:
utils.report_error(
"CLASSPATH is not configured with the path to picard.jar")
found_all_dependencies = False
else:
stdout = command.run("java -jar " + jar_file_path + " 2>&1")
if stdout.lower().startswith("error"):
utils.report_error(stdout)
found_all_dependencies = False
gatk_required = os.environ["EnableLocalRealignment"] == "true"
if gatk_required:
jar_file_path = utils.find_path_in_path_list("GenomeAnalysisTK",
"CLASSPATH")
if not jar_file_path:
utils.report_error(
"CLASSPATH is not configured with the path to GenomeAnalysisTK.jar"
)
found_all_dependencies = False
else:
stdout = command.run("java -jar " + jar_file_path +
" --version 2>&1")
if stdout.lower().startswith("error"):
utils.report_error(stdout)
found_all_dependencies = False
else:
stdout = command.run("java -jar " + jar_file_path +
" -T IndelRealigner --version 2>&1")
if "not a valid command" in stdout.lower(
) or "indelrealigner is no longer included" in stdout.lower():
utils.report_error(
"The installed GATK version does not support indel realignment. Try installing an older release prior to GATK v4."
)
found_all_dependencies = False
elif "user error has occurred" in stdout.lower():
utils.report_error(stdout)
found_all_dependencies = False
if not found_all_dependencies:
utils.fatal_error(
"Check the SNP Pipeline installation instructions here: http://snp-pipeline.readthedocs.org/en/latest/installation.html"
)
else:
print("OK")
# Process the sample directory command line option
# TODO: detect broken fastq symlinks
if args.samplesDir:
samples_parent_dir = args.samplesDir.rstrip(
'/') # strip trailing slash
if not utils.verify_non_empty_directory("Samples directory",
samples_parent_dir):
sys.exit(1)
# verify at least one of the subdirectories contains fastq files.
dir_sizes = get_sorted_sample_dirs_fastq_sizes(samples_parent_dir)
dir_sizes = [(size, path) for size, path in dir_sizes if size > 0]
if len(dir_sizes) == 0:
utils.fatal_error(
"Samples directory %s does not contain subdirectories with fastq files."
% samples_parent_dir)
sample_dirs_file = os.path.join(work_dir, "sampleDirectories.txt")
persist_sorted_sample_dirs_file(samples_parent_dir, sample_dirs_file)
# Process the file of sample directories command line option
# TODO: detect broken fastq symlinks
if args.samplesFile:
sample_dirs_file = args.samplesFile
if not os.path.isfile(sample_dirs_file):
utils.fatal_error(
"Error: the file of samples directories, %s, does not exist." %
sample_dirs_file)
if os.path.getsize(sample_dirs_file) == 0:
utils.fatal_error(
"Error: the file of samples directories, %s, is empty." %
sample_dirs_file)
rewrite_cleansed_file_of_sample_dirs(
sample_dirs_file, os.path.join(work_dir, "sampleDirectories.txt"))
sample_dirs_file = os.path.join(work_dir, "sampleDirectories.txt")
validate_file_of_sample_dirs(sample_dirs_file)
with open(sample_dirs_file) as f:
sample_dirs_list = f.read().splitlines()
sample_count = len(sample_dirs_list)
# --------------------------------------------------------
if job_queue_mgr is None:
progress("Step 1 - Prep work")
else:
print("Step 1 - Prep work")
# --------------------------------------------------------
# Mirror the input reference and samples if requested
# TODO: make this a pure python solution
if args.mirror:
if args.mirror == "soft":
# soft link, subsequent freshness checks use the timestamp of original file, not the soft link
mirror_flag = " -s "
elif args.mirror == "hard":
# hard link, automatically preserves attributes of the original file
mirror_flag = " -l "
else:
# regular copy, -p explicitly preserves attributes of the original file
mirror_flag = " -p "
# flush stdout to keep the unbuffered stderr in chronological order with stdout
sys.stdout.flush()
# Mirror/link the reference
work_reference_dir = os.path.join(work_dir, "reference")
utils.mkdir_p(work_reference_dir)
src_reference_file = os.path.abspath(reference_file_path)
cmd = "cp -v -u -f" + mirror_flag + src_reference_file + ' ' + work_reference_dir
subprocess.check_call(cmd, shell=True)
# since we mirrored the reference, we need to update our reference location
reference_file_path = os.path.join(work_reference_dir,
reference_file_name)
# Mirror/link the samples
work_samples_parent_dir = os.path.join(work_dir, "samples")
for directory in sample_dirs_list:
basedir = os.path.basename(directory)
work_sample_dir = os.path.join(work_samples_parent_dir, basedir)
utils.mkdir_p(work_sample_dir)
src_sample_dir = os.path.abspath(directory)
# copy without stderr message and without exit error code because the fastq or fq files might not exist
cmd = "cp -r -v -u -f" + mirror_flag + src_sample_dir + "/*.fastq* " + work_sample_dir + " 2> /dev/null || true"
subprocess.check_call(cmd, shell=True)
cmd = "cp -r -v -u -f" + mirror_flag + src_sample_dir + "/*.fq* " + work_sample_dir + " 2> /dev/null || true"
subprocess.check_call(cmd, shell=True)
# since we mirrored the samples, we need to update our sorted list of samples
sample_dirs_file = os.path.join(work_dir, "sampleDirectories.txt")
persist_sorted_sample_dirs_file(work_samples_parent_dir,
sample_dirs_file)
# refresh the list of sample dirs -- now in sorted order
with open(sample_dirs_file) as f:
sample_dirs_list = f.read().splitlines()
# get the *.fastq or *.fq files in each sample directory, possibly compresessed, on one line per sample, ready to feed to bowtie
sample_full_path_names_file = os.path.join(work_dir,
"sampleFullPathNames.txt")
with open(sample_full_path_names_file, 'w') as f:
for directory in sample_dirs_list:
file_list = fastq.list_fastq_files(directory)
print(' '.join(file_list), file=f)
# Initialize the job runner
if job_queue_mgr is None:
runner = JobRunner("local",
exception_handler=handle_exception,
verbose=args.verbose >= 4)
elif job_queue_mgr == "grid":
strip_job_array_suffix = config_params.get(
"GridEngine_StripJobArraySuffix", "true").lower()
qsub_extra_params = config_params.get("GridEngine_QsubExtraParams")
runner = JobRunner(job_queue_mgr,
strip_job_array_suffix == "true",
qsub_extra_params=qsub_extra_params,
verbose=args.verbose >= 4)
else:
strip_job_array_suffix = config_params.get(
"Torque_StripJobArraySuffix", "false").lower()
qsub_extra_params = config_params.get("Torque_QsubExtraParams")
runner = JobRunner(job_queue_mgr,
strip_job_array_suffix == "true",
qsub_extra_params=qsub_extra_params,
verbose=args.verbose >= 4)
progress("Step 2 - Index the reference")
log_file = os.path.join(log_dir, "indexRef.log")
command_line = "cfsan_snp_pipeline index_ref" + force_flag + reference_file_path
job_id_index_ref = runner.run(command_line, "indexRef", log_file)
progress("Step 3 - Map the sample reads to the reference")
# Parse the user-specified aligner parameters to find the number of CPU cores requested, for example, "-p 16" or "-n 16"
# Set the default number of CPU cores if the user did not configure a value.
if snp_pipeline_aligner == "smalt":
extra_params_env_var = "SmaltAlign_ExtraParams"
threads_option = "-n"
else:
extra_params_env_var = "Bowtie2Align_ExtraParams"
threads_option = "-p"
aligner_max_processes, aligner_threads_per_process = utils.configure_process_threads(
extra_params_env_var, threads_option, cpu_cores_per_process,
max_cpu_cores)
samfilter_max_processes, samfilter_threads_per_process = utils.configure_process_threads(
"SamtoolsSamFilter_ExtraParams", ["-@", "--threads"],
cpu_cores_per_process, max_cpu_cores)
samsort_max_processes, samsort_threads_per_process = utils.configure_process_threads(
"SamtoolsSort_ExtraParams", ["-@", "--threads"], cpu_cores_per_process,
max_cpu_cores)
samindex_max_processes, samindex_threads_per_process = utils.configure_process_threads(
"SamtoolsIndex_ExtraParams", ["-@"], cpu_cores_per_process,
max_cpu_cores)
realigner_max_processes, realigner_threads_per_process = utils.configure_process_threads(
"RealignerTargetCreator_ExtraParams", ["-nt", "--num_threads"],
cpu_cores_per_process, max_cpu_cores)
# There are multiple processes within map_reads, each with multiple threads.
# The CPU allocation must be enough for the process needing the largest number of threads.
max_processes_list = [
aligner_max_processes, samfilter_max_processes, samsort_max_processes,
samindex_max_processes, realigner_max_processes
]
if all([i is None for i in max_processes_list]):
max_processes = None
else:
max_processes = min([i for i in max_processes_list if i is not None])
threads_per_process = max(aligner_threads_per_process,
samfilter_threads_per_process,
samsort_threads_per_process,
samindex_threads_per_process,
realigner_threads_per_process)
parallel_environment = config_params.get("GridEngine_PEname", None)
log_file = os.path.join(log_dir, "mapReads.log")
command_line = "cfsan_snp_pipeline map_reads --threads " + str(
threads_per_process) + force_flag + reference_file_path + " {1} {2}"
job_id_map_reads = runner.run_array(
command_line,
"mapReads",
log_file,
sample_full_path_names_file,
max_processes=max_processes,
wait_for=[job_id_index_ref],
threads=threads_per_process,
parallel_environment=parallel_environment)
progress("Step 4 - Find sites with SNPs in each sample")
if job_queue_mgr in ["grid", "torque"]:
time.sleep(
1.0 + float(sample_count) / 150
) # workaround torque bug when submitting two large consecutive array jobs, potential bug for grid
log_file = os.path.join(log_dir, "callSites.log")
command_line = "cfsan_snp_pipeline call_sites" + force_flag + reference_file_path + " {1}"
job_id_call_sites = runner.run_array(command_line,
"callSites",
log_file,
sample_dirs_file,
max_processes=max_cpu_cores,
wait_for_array=[job_id_map_reads],
slot_dependency=True)
progress("Step 5 - Filter abnormal SNP regions")
log_file = os.path.join(log_dir, "filterRegions.log")
extra_params = os.environ.get("FilterRegions_ExtraParams", "")
command_line = "cfsan_snp_pipeline filter_regions" + force_flag + "-n var.flt.vcf " + sample_dirs_file + ' ' + reference_file_path + ' ' + extra_params
job_id_filter_regions = runner.run(command_line,
"filterRegions",
log_file,
wait_for_array=[job_id_call_sites])
# Starting from here, there are 2 threads:
# Thread X.1: the thread processing the original VCF files and corresponding downstream results
# Thread X.2: the thread processing the preserved VCF files and corresponding downstream results
progress(
"Step 6.1 - Merge the SNP sites across all samples into the SNP list file"
)
# The mergeSites process creates the filtered list of sample directories. It is the list of samples not having excessive snps.
# When running on a workstation, the file exists at this point during the script execution, but on grid or torque, it has not yet been created. However,
# we know the path to the file regardless of whether it exists yet.
filtered_sample_dirs_file = sample_dirs_file + ".OrigVCF.filtered"
# touch $filtered_sample_dirs_file # TODO: why was this touch here in the old run_snp_pipeline.sh script?
log_file = os.path.join(log_dir, "mergeSites.log")
output_file = os.path.join(work_dir, "snplist.txt")
extra_params = os.environ.get("MergeSites_ExtraParams", "")
command_line = "cfsan_snp_pipeline merge_sites" + force_flag + "-n var.flt.vcf -o " + output_file + ' ' + extra_params + ' ' + sample_dirs_file + ' ' + filtered_sample_dirs_file
job_id_merge_sites = runner.run(command_line,
"mergeSites",
log_file,
wait_for=[job_id_filter_regions])
progress(
"Step 6.2 - Merge the SNP sites across all samples into the SNP list file"
)
# Create another copy of sample directories file, for the thread processing preserved snp files.
filtered_sample_dirs_file2 = sample_dirs_file + ".PresVCF.filtered"
# touch $filtered_sample_dirs_file2 # TODO: why was this touch here in the old run_snp_pipeline.sh script?
log_file = os.path.join(log_dir, "mergeSites_preserved.log")
output_file = os.path.join(work_dir, "snplist_preserved.txt")
extra_params = os.environ.get("MergeSites_ExtraParams", "")
command_line = "cfsan_snp_pipeline merge_sites" + force_flag + "-n var.flt_preserved.vcf -o " + output_file + ' ' + extra_params + ' ' + sample_dirs_file + ' ' + filtered_sample_dirs_file2
job_id_merge_sites2 = runner.run(command_line,
"mergeSites_preserved",
log_file,
wait_for=[job_id_filter_regions])
progress("Step 7.1 - Call the consensus SNPs for each sample")
log_file = os.path.join(log_dir, "callConsensus.log")
list_file = os.path.join(work_dir, "snplist.txt")
output_file = "{1}/consensus.fasta"
extra_params = os.environ.get("CallConsensus_ExtraParams", "")
command_line = "cfsan_snp_pipeline call_consensus" + force_flag + "-l " + list_file + " -o " + output_file + " --vcfRefName " + reference_file_name + ' ' + extra_params + " --vcfFileName consensus.vcf {1}/reads.all.pileup"
job_id_call_consensus = runner.run_array(command_line,
"callConsensus",
log_file,
sample_dirs_file,
max_processes=max_cpu_cores,
wait_for=[job_id_merge_sites])
progress("Step 7.2 - Call the consensus SNPs for each sample")
log_file = os.path.join(log_dir, "callConsensus_preserved.log")
list_file = os.path.join(work_dir, "snplist_preserved.txt")
output_file = "{1}/consensus_preserved.fasta"
extra_params = os.environ.get("CallConsensus_ExtraParams", "")
command_line = "cfsan_snp_pipeline call_consensus" + force_flag + "-l " + list_file + " -o " + output_file + " -e {1}/var.flt_removed.vcf --vcfRefName " + reference_file_name + ' ' + extra_params + " --vcfFileName consensus_preserved.vcf {1}/reads.all.pileup"
job_id_call_consensus2 = runner.run_array(command_line,
"callConsensus_preserved",
log_file,
sample_dirs_file,
max_processes=max_cpu_cores,
wait_for=[job_id_merge_sites2])
progress("Step 8.1 - Create the SNP matrix")
log_file = os.path.join(log_dir, "snpMatrix.log")
output_file = os.path.join(work_dir, "snpma.fasta")
extra_params = os.environ.get("SnpMatrix_ExtraParams", "")
command_line = "cfsan_snp_pipeline snp_matrix" + force_flag + "-c consensus.fasta -o " + output_file + ' ' + extra_params + ' ' + filtered_sample_dirs_file
job_id_snp_matrix = runner.run(command_line,
"snpMatrix",
log_file,
wait_for_array=[job_id_call_consensus])
progress("Step 8.2 - Create the SNP matrix")
log_file = os.path.join(log_dir, "snpMatrix_preserved.log")
output_file = os.path.join(work_dir, "snpma_preserved.fasta")
extra_params = os.environ.get("SnpMatrix_ExtraParams", "")
command_line = "cfsan_snp_pipeline snp_matrix" + force_flag + "-c consensus_preserved.fasta -o " + output_file + ' ' + extra_params + ' ' + filtered_sample_dirs_file2
job_id_snp_matrix2 = runner.run(command_line,
"snpMatrix_preserved",
log_file,
wait_for_array=[job_id_call_consensus2])
progress("Step 9.1 - Create the reference sequence at SNP sites")
log_file = os.path.join(log_dir, "snpReference.log")
list_file = os.path.join(work_dir, "snplist.txt")
output_file = os.path.join(work_dir, "referenceSNP.fasta")
extra_params = os.environ.get("SnpReference_ExtraParams", "")
command_line = "cfsan_snp_pipeline snp_reference" + force_flag + "-l " + list_file + " -o " + output_file + ' ' + extra_params + ' ' + reference_file_path
job_id_snp_reference = runner.run(command_line,
"snpReference",
log_file,
wait_for_array=[job_id_call_consensus])
progress("Step 9.2 - Create the reference sequence at SNP sites")
log_file = os.path.join(log_dir, "snpReference_preserved.log")
list_file = os.path.join(work_dir, "snplist_preserved.txt")
output_file = os.path.join(work_dir, "referenceSNP_preserved.fasta")
extra_params = os.environ.get("SnpReference_ExtraParams", "")
command_line = "cfsan_snp_pipeline snp_reference" + force_flag + "-l " + list_file + " -o " + output_file + ' ' + extra_params + ' ' + reference_file_path
job_id_snp_reference2 = runner.run(command_line,
"snpReference_preserved",
log_file,
wait_for_array=[job_id_call_consensus2])
progress("Step 10.1 - Merge sample VCFs to create the multi-VCF file")
if "--vcfFileName" in os.environ.get("CallConsensus_ExtraParams", ""):
log_file = os.path.join(log_dir, "mergeVcfs.log")
output_file = os.path.join(work_dir, "snpma.vcf")
extra_params = os.environ.get("MergeVcfs_ExtraParams", "")
command_line = "cfsan_snp_pipeline merge_vcfs" + force_flag + "-o " + output_file + ' ' + extra_params + ' ' + filtered_sample_dirs_file
job_id_merge_vcfs = runner.run(command_line,
"mergeVcfs",
log_file,
wait_for_array=[job_id_call_consensus])
else:
print("Skipped per CallConsensus_ExtraParams configuration")
progress("Step 10.2 - Merge sample VCFs to create the multi-VCF file")
if "--vcfFileName" in os.environ.get("CallConsensus_ExtraParams", ""):
log_file = os.path.join(log_dir, "mergeVcfs_preserved.log")
output_file = os.path.join(work_dir, "snpma_preserved.vcf")
extra_params = os.environ.get("MergeVcfs_ExtraParams", "")
command_line = "cfsan_snp_pipeline merge_vcfs" + force_flag + "-n consensus_preserved.vcf -o " + output_file + ' ' + extra_params + ' ' + filtered_sample_dirs_file2
job_id_merge_vcfs2 = runner.run(
command_line,
"mergeVcfs_preserved",
log_file,
wait_for_array=[job_id_call_consensus2])
else:
print("Skipped per CallConsensus_ExtraParams configuration")
progress("Step 11.1 - Calculate SNP distance matrix")
log_file = os.path.join(log_dir, "distance.log")
input_file = os.path.join(work_dir, "snpma.fasta")
pair_output_file = os.path.join(work_dir, "snp_distance_pairwise.tsv")
matrix_output_file = os.path.join(work_dir, "snp_distance_matrix.tsv")
command_line = "cfsan_snp_pipeline distance" + force_flag + "-p " + pair_output_file + " -m " + matrix_output_file + ' ' + input_file
job_id_distance = runner.run(command_line,
"distance",
log_file,
wait_for=[job_id_snp_matrix])
progress("Step 11.2 - Calculate SNP distance matrix")
log_file = os.path.join(log_dir, "distance_preserved.log")
input_file = os.path.join(work_dir, "snpma_preserved.fasta")
pair_output_file = os.path.join(work_dir,
"snp_distance_pairwise_preserved.tsv")
matrix_output_file = os.path.join(work_dir,
"snp_distance_matrix_preserved.tsv")
command_line = "cfsan_snp_pipeline distance" + force_flag + "-p " + pair_output_file + " -m " + matrix_output_file + ' ' + input_file
job_id_distance2 = runner.run(command_line,
"distance_preserved",
log_file,
wait_for=[job_id_snp_matrix2])
progress("Step 12 - Collect metrics for each sample")
log_file = os.path.join(log_dir, "collectMetrics.log")
output_file = "{1}/metrics"
extra_params = os.environ.get("CollectMetrics_ExtraParams", "")
command_line = "cfsan_snp_pipeline collect_metrics" + force_flag + "-o " + output_file + ' ' + extra_params + " {1} " + reference_file_path
job_id_collect_metrics = runner.run_array(
command_line,
"collectMetrics",
log_file,
sample_dirs_file,
max_processes=max_cpu_cores,
wait_for_array=[job_id_call_consensus, job_id_call_consensus2],
slot_dependency=True)
progress(
"Step 13 - Combine the metrics across all samples into the metrics table"
)
log_file = os.path.join(log_dir, "combineMetrics.log")
output_file = os.path.join(work_dir, "metrics.tsv")
extra_params = os.environ.get("CombineMetrics_ExtraParams", "")
command_line = "cfsan_snp_pipeline combine_metrics" + force_flag + "-n metrics -o " + output_file + ' ' + extra_params + ' ' + sample_dirs_file
combine_metrics_job_id = runner.run(
command_line,
"combineMetrics",
log_file,
wait_for_array=[job_id_collect_metrics])
# Decide whether to purge the intermediate output files upon successful completion.
# Case 1: we are running on the HPC. We always need to submit the purge task. It will decide to do nothing if there were errors.
if job_queue_mgr is not None: # HPC
need_purge = args.purge # need to submit the purge task, it might decide to do nothing if there were errors
# Case 2: we are running locally and we know right now whether there were any errors.
# Case 2a: We are configured to stop on error, but the fact that we got this far means there were no errors -- so we need to purge.
# Case 2b: We are configured to ignore errors, so now we look for evidence of errors and purge if there were no errors.
else:
errors_detected = os.path.isfile(error_output_file)
need_purge = args.purge and not errors_detected
if need_purge:
progress("Step 14 - Purge the intermediate output files")
log_file = os.path.join(log_dir, "purge.log")
command_line = "cfsan_snp_pipeline purge " + work_dir
purge_job_id = runner.run(command_line,
"purge",
log_file,
wait_for=[combine_metrics_job_id])
# Step 15 - Notify user of any non-fatal errors accumulated during processing
if os.path.isfile(error_output_file) and os.path.getsize(
error_output_file) > 0 and not stop_on_error:
print(
"\nThere were errors processing some samples.\nSee the log file %s for a summary of errors."
% error_output_file,
file=sys.stderr)
# Exit here to prevent showing the "cfsan_snp_pipeline run finished" message. The jobs are queued, not finished yet.
if job_queue_mgr is not None: # HPC
sys.exit(0)
else:
end_time = time.time()
elapsed_time = end_time - start_time
print("Elapsed time =", elapsed_time)