def collect_metrics(args):
"""Collect the quality metrics and SNP metrics for a sample.
This function expects, or creates '(*)', the following files arranged
in the following way:
reference
referenceFile.fasta
samples
sample_name_one/*.fastq.gz
sample_name_one/reads.sam
sample_name_one/reads.sorted.deduped.bam
sample_name_one/reads.sorted.bam
sample_name_one/reads.all.pileup
sample_name_one/var.flt.vcf
sample_name_one/var.flt_preserved.vcf
sample_name_one/consensus.fasta
sample_name_one/consensus_preserved.fasta
sample_name_one/consensus.vcf
sample_name_one/consensus_preserved.vcf
sample_name_one/metrics*
The input files are created outside of this function. The package
documentation provides an example of preparing these files based on the
lambda_virus sequence that is used as one test for this package.
Parameters
----------
args : argparse.Namespace
referenceFile : File path of the reference fasta file
sampleDir : Relative or absolute directory of the sample
consensusFastaFileName : File name of the consensus fasta file which must exist in the sample directory
consensusPreservedFastaFileName : File name of the consensus preserved fasta file which must exist in the sample directory
consensusVcfFileName : File name of the consensus vcf file which must exist in the sample directory
consensusPreservedVcfFileName : File name of the consensus preserved vcf file which must exist in the sample directory
maxSnps : Maximum allowed number of SNPs per sample
metricsFile : Output file. Relative or absolute path to the metrics file
"""
utils.print_log_header(classpath=True)
utils.print_arguments(args)
#==========================================================================
# Validate inputs
#==========================================================================
# Verify reference fasta file exists and is not empty
reference_file_path = args.referenceFile
utils.verify_non_empty_input_files("Reference file", [reference_file_path], error_handler="global")
sample_dir = args.sampleDir
utils.verify_non_empty_directory("Sample directory", sample_dir, error_handler="sample", continue_possible=False)
metrics_file_path = args.metricsFile
max_allowed_snps = args.maxSnps
consensus_vcf_file_name = args.consensusVcfFileName
consensus_preserved_vcf_file_name = args.consensusPreservedVcfFileName
consensus_fasta_file_name = args.consensusFastaFileName
consensus_preserved_fasta_file_name = args.consensusPreservedFastaFileName
sample_id = utils.sample_id_from_dir(sample_dir)
#==========================================================================
# Read existing metrics file so some metrics can be reused
#==========================================================================
try:
metrics = utils.read_properties(metrics_file_path)
except IOError:
metrics = dict()
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(), "Get machine and flowcell from fastq header"))
#-------------------------
machine = ""
flowcell = ""
fastq_files = fastq.list_fastq_files(sample_dir)
fastq_files = [f for f in fastq_files if os.path.isfile(f)] # Exclude broken symlinks
if not fastq_files:
handle_error("No fastq files were found.")
else:
tags = fastq.extract_metadata_tags(fastq_files[0])
if tags:
machine = tags.instrument or ""
flowcell = tags.flow_cell or ""
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(), "Sum file sizes of paired fastq files"))
#-------------------------
fastq_file_size = ""
fastq_file_list = ""
if fastq_files:
fastq_file_size = sum([os.path.getsize(file) for file in fastq_files])
# Make a comma separated list of just the fastq file names without directories
fastq_file_list = [os.path.basename(file) for file in fastq_files]
fastq_file_list = ", ".join(fastq_file_list)
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(), "Calculate number of reads and %mapped from sam file"))
#-------------------------
num_reads = ""
percent_reads_mapped = ""
file = os.path.join(sample_dir, "reads.sam")
if verify_input_file("SAM file", file):
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
num_reads = metrics.get("numberReads", "") # reuse already fresh metrics
percent_reads_mapped = metrics.get("percentReadsMapped", "") # reuse already fresh metrics
if num_reads and percent_reads_mapped:
verbose_print("Reusing previously calculated number of reads and %mapped")
else:
num_reads = command.run("samtools view -S -c " + file)
num_reads = num_reads.strip()
mapped = command.run("samtools view -S -c -F 4 " + file)
mapped = mapped.strip()
try:
percent_reads_mapped = 100.0 * float(mapped) / float(num_reads)
percent_reads_mapped = "%.2f" % percent_reads_mapped
except ValueError:
handle_error("Cannot calculate number of reads and %mapped.")
#-------------------------
# Calculate number of duplicate reads from deduped bam file
#-------------------------
num_dup_reads = ""
remove_duplicate_reads = os.environ.get("RemoveDuplicateReads") or "true"
remove_duplicate_reads = remove_duplicate_reads.lower()
if remove_duplicate_reads == "true":
verbose_print("# %s %s" % (utils.timestamp(), "Calculate number of duplicate reads from deduped bam file"))
file = os.path.join(sample_dir, "reads.sorted.deduped.bam")
if verify_input_file("Deduped BAM file", file):
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
num_dup_reads = metrics.get("numberDupReads", "") # reuse already fresh metrics
if num_dup_reads:
verbose_print("Reusing previously calculated number of duplicate reads")
else:
num_dup_reads = command.run("samtools view -S -c -f 1024 " + file)
num_dup_reads = num_dup_reads.strip()
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(), "Calculate mean insert size from bam file"))
#-------------------------
ave_insert_size = ""
file = os.path.join(sample_dir, "reads.sorted.bam")
if verify_input_file("BAM file", file):
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
ave_insert_size = metrics.get("aveInsertSize", "") # reuse already fresh metrics
if ave_insert_size:
verbose_print("Reusing previously calculated mean insert size")
else:
# Extract inferred insert sizes (TLEN, column 9 of BAM file) for reads "mapped in proper pair" (2) and "first in pair" (64) = 66
tempfile = NamedTemporaryFile(delete=False, dir=sample_dir, prefix="tmp.inserts.", mode='w')
command.run("samtools view -f 66 " + file + " | cut -f 9 | sed 's/^-//'", tempfile.name)
insert_count = 0
insert_sum = 0
with open(tempfile.name) as f:
for line in f:
try:
insert_sum += int(line)
insert_count += 1
except ValueError:
pass
os.unlink(tempfile.name)
if insert_count > 0 and insert_sum > 0:
ave_insert_size = float(insert_sum) / float(insert_count)
ave_insert_size = "%.2f" % ave_insert_size
else:
handle_error("Cannot calculate mean insert size.")
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(), "Calculate mean depth from pileup file"))
#-------------------------
ave_pileup_depth = ""
file = os.path.join(sample_dir, "reads.all.pileup")
if verify_input_file("Pileup file", file):
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
ave_pileup_depth = metrics.get("avePileupDepth", "") # reuse already fresh metrics
if ave_pileup_depth:
verbose_print("Reusing previously calculated mean pileup depth")
else:
depth_sum = 0
with open(file) as f:
for line in f:
tokens = line.split()
try:
depth_sum += int(tokens[3])
except (ValueError, IndexError):
pass
reference_length = 0
for record in SeqIO.parse(reference_file_path, "fasta"):
reference_length += len(record)
if depth_sum > 0 and reference_length > 0:
#print("depth_sum=%i" % depth_sum);
#print("reference_length=%i" % reference_length)
ave_pileup_depth = float(depth_sum) / float(reference_length)
ave_pileup_depth = "%.2f" % ave_pileup_depth
else:
handle_error("Cannot calculate mean pileup depth.")
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(), "Count number of high confidence SNP positions from phase 1 vcf file"))
#-------------------------
phase1_snps = ""
excluded_sample = ""
file = os.path.join(sample_dir, "var.flt.vcf")
if verify_input_file("VCF file", file):
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
phase1_snps = metrics.get("phase1Snps", "") # reuse already fresh metrics
if phase1_snps:
verbose_print("Reusing previously calculated phase1 snps")
else:
phase1_snps = count_vcf_file_snps(file)
# Flag excessive snps
if max_allowed_snps > 0 and phase1_snps > max_allowed_snps:
excluded_sample = "Excluded"
handle_error("Excluded: exceeded %i maxsnps." % max_allowed_snps)
phase1_snps = str(phase1_snps)
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(), "Count number of filter_regions preserved high confidence SNP positions from phase 1 vcf file"))
#-------------------------
phase1_snps_preserved = ""
excluded_sample_preserved = ""
file = os.path.join(sample_dir, "var.flt_preserved.vcf")
if verify_input_file("VCF file", file):
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
phase1_snps_preserved = metrics.get("phase1SnpsPreserved", "") # reuse already fresh metrics
if phase1_snps_preserved:
verbose_print("Reusing previously calculated preserved phase1 snps")
else:
phase1_snps_preserved = count_vcf_file_snps(file)
# Flag excessive snps
if max_allowed_snps > 0 and phase1_snps_preserved > max_allowed_snps:
excluded_sample_preserved = "Excluded"
handle_error("Excluded: preserved exceeded %i maxsnps." % max_allowed_snps)
phase1_snps_preserved = str(phase1_snps_preserved)
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(), "Count number of consensus snps from consensus vcf file"))
#-------------------------
phase2_snps = ""
file = os.path.join(sample_dir, consensus_vcf_file_name)
if verify_input_file("Consensus VCF file", file):
# Omit the phase2 snp count if the sample is excluded.
# It will be meaningless since this sample's phase1 snps are excluded from the snplist.
if excluded_sample != "Excluded":
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
phase2_snps = metrics.get("snps", "") # reuse already fresh metrics
if phase2_snps:
verbose_print("Reusing previously calculated phase2 snps")
else:
phase2_snps = count_vcf_file_snps(file)
phase2_snps = str(phase2_snps)
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(), "Count number of preserved consensus snps from consensus vcf file"))
#-------------------------
phase2_snps_preserved = ""
file = os.path.join(sample_dir, consensus_preserved_vcf_file_name)
if verify_input_file("Consensus VCF file", file):
# Omit the phase2 snp count if the sample is excluded.
# It will be meaningless since this sample's phase1 snps are excluded from the snplist.
if excluded_sample_preserved != "Excluded":
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
phase2_snps_preserved = metrics.get("snpsPreserved", "") # reuse already fresh metrics
if phase2_snps_preserved:
verbose_print("Reusing previously calculated preserved phase2 snps")
else:
phase2_snps_preserved = count_vcf_file_snps(file)
phase2_snps_preserved = str(phase2_snps_preserved)
#------------------------------------------
verbose_print("# %s %s" % (utils.timestamp(), "Count missing positions in the snp matrix"))
#------------------------------------------
missing_pos = ""
file = os.path.join(sample_dir, consensus_fasta_file_name)
if verify_input_file("Consensus fasta file", file):
# Omit the phase2 gap count if the sample is excluded.
# It will be meaningless since this sample's phase1 snps are excluded from the snplist.
if excluded_sample != "Excluded":
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
missing_pos = metrics.get("missingPos", "") # reuse already fresh metrics
if missing_pos:
verbose_print("Reusing previously calculated missing positions")
else:
missing_pos = count_missing_snp_matrix_positions(file, sample_id)
missing_pos = str(missing_pos)
#------------------------------------------
verbose_print("# %s %s" % (utils.timestamp(), "Count missing positions in the preserved snp matrix"))
#------------------------------------------
missing_pos_preserved = ""
file = os.path.join(sample_dir, consensus_preserved_fasta_file_name)
if verify_input_file("Consensus fasta file", file):
# Omit the phase2 gap count if the sample is excluded.
# It will be meaningless since this sample's phase1 snps are excluded from the snplist.
if excluded_sample_preserved != "Excluded":
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
missing_pos_preserved = metrics.get("missingPosPreserved", "") # reuse already fresh metrics
if missing_pos_preserved:
verbose_print("Reusing previously calculated missing positions")
else:
missing_pos_preserved = count_missing_snp_matrix_positions(file, sample_id)
missing_pos_preserved = str(missing_pos_preserved)
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(), "Print results"))
#-------------------------
with open(metrics_file_path, "w") as f:
print("sample=" + '"' + sample_id + '"', file=f)
print("fastqFileList=" + '"' + fastq_file_list + '"', file=f)
print("fastqFileSize=" + str(fastq_file_size), file=f)
print("machine=" + machine, file=f)
print("flowcell=" + flowcell, file=f)
print("numberReads=" + num_reads, file=f)
print("numberDupReads=" + num_dup_reads, file=f)
print("percentReadsMapped=" + percent_reads_mapped, file=f)
print("aveInsertSize=" + ave_insert_size, file=f)
print("avePileupDepth=" + ave_pileup_depth, file=f)
print("phase1Snps=" + phase1_snps, file=f)
print("phase1SnpsPreserved=" + phase1_snps_preserved, file=f)
print("snps=" + phase2_snps, file=f)
print("snpsPreserved=" + phase2_snps_preserved, file=f)
print("missingPos=" + missing_pos, file=f)
print("missingPosPreserved=" + missing_pos_preserved, file=f)
print("excludedSample=" + excluded_sample, file=f)
print("excludedSamplePreserved=" + excluded_sample_preserved, file=f)
print("errorList=" + '"' + ' '.join(error_list) + '"', file=f)
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
# 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
# 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["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["RemoveDuplicateReads"] = config_params.get("RemoveDuplicateReads", "").lower() or "true"
os.environ["PicardMarkDuplicates_ExtraParams"] = config_params.get("PicardMarkDuplicates_ExtraParams", "")
os.environ["PicardJvm_ExtraParams"] = config_params.get("PicardJvm_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
dependencies = ["cfsan_snp_pipeline", snp_pipeline_aligner, "samtools", "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
stdout = command.run("java net.sf.varscan.VarScan 2>&1")
if "Error" in stdout:
utils.report_error("CLASSPATH is not configured with the path to VarScan")
found_all_dependencies = False
if os.environ["RemoveDuplicateReads"] == "true":
stdout = command.run("java picard.cmdline.PicardCommandLine 2>&1")
if "Error" in stdout:
utils.report_error("CLASSPATH is not configured with the path to Picard")
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")
# 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"
max_processes, threads_per_process = configure_process_threads(extra_params_env_var, threads_option, 8, max_cpu_cores)
parallel_environment = config_params.get("GridEngine_PEname", None)
log_file = os.path.join(log_dir, "mapReads.log")
command_line = "cfsan_snp_pipeline map_reads" + 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)
def combine_metrics(args):
"""Combine the per-sample metrics files into a single table of metrics for all samples.
This function expects, or creates '(*)', the following files arranged
in the following way:
samples
sample_name_one/metrics
metrics.tsv
All the input files are created outside of this function. Before
running this command, the metrics file for each sample must be created by the
collect_metrics command.
The package documentation provides an example of preparing these files based
on the lambda_virus sequence that is used as one test for this package.
Parameters
----------
args : argparse.Namespace
sampleDirsFile : Path to file containing a list of directories -- one per sample
metricsFileName : File name of the metrics files which must exist in each of the sample directories
mergedMetricsFile : Path to the output merged metrics file
"""
utils.print_log_header()
utils.print_arguments(args)
#==========================================================================
# Validate inputs
#==========================================================================
sample_directories_list_path = args.sampleDirsFile
utils.verify_non_empty_input_files("File of sample directories", [sample_directories_list_path], error_handler="global")
metrics_file_name = args.metricsFileName
merged_metrics_path = args.mergedMetricsFile
with open(sample_directories_list_path, "r") as f:
sample_directories = [line.rstrip() for line in f]
sample_directories = [d for d in sample_directories if d]
metrics_files = [os.path.join(d, metrics_file_name) for d in sample_directories]
#==========================================================================
# Check if merge has already been done
#==========================================================================
needs_rebuild = utils.target_needs_rebuild(metrics_files, merged_metrics_path)
if not args.forceFlag and not needs_rebuild:
verbose_print("# The merged metrics file is already freshly created. Use the -f option to force a rebuild.")
return
#==========================================================================
# Parse the metrics files and print the tabular results
#==========================================================================
with open(merged_metrics_path, 'w') as f:
# Emit the column headings
column_headings = ["Sample", "Fastq Files", "Fastq File Size", "Machine", "Flowcell", "Number of Reads", "Duplicate Reads", "Percent of Reads Mapped",
"Percent Proper Pair",
"Average Insert Size", "Average Pileup Depth", "Phase1 SNPs", "Phase1 Preserved SNPs", "Phase2 SNPs", "Phase2 Preserved SNPs",
"Missing SNP Matrix Positions", "Missing Preserved SNP Matrix Positions", "Excluded Sample", "Excluded Preserved Sample", "Warnings and Errors"]
if not args.spaceHeadings:
column_headings = [heading.replace(' ', '_') for heading in column_headings]
tabbed_headings = '\t'.join(column_headings)
f.write(tabbed_headings + '\n')
# Reads the metrics from each sample, and emit the values
for metrics_file in metrics_files:
verbose_print("Processing " + metrics_file)
message = None
if not os.path.isfile(metrics_file):
message = "Sample metrics file %s does not exist." % metrics_file
elif os.path.getsize(metrics_file) == 0:
message = "Sample metrics file %s is empty." % metrics_file
if message:
f.write(message + '\n')
utils.sample_warning(message)
continue
metrics = utils.read_properties(metrics_file)
f.write(quoted(metrics.get("sample", "")) + '\t')
f.write(quoted(metrics.get("fastqFileList", "")) + '\t')
f.write(metrics.get("fastqFileSize", "") + '\t')
f.write(metrics.get("machine", "") + '\t')
f.write(metrics.get("flowcell", "") + '\t')
f.write(metrics.get("numberReads", "") + '\t')
f.write(metrics.get("numberDupReads", "") + '\t')
f.write(metrics.get("percentReadsMapped", "") + '\t')
f.write(metrics.get("percentProperPair", "") + '\t')
f.write(metrics.get("aveInsertSize", "") + '\t')
f.write(metrics.get("avePileupDepth", "") + '\t')
f.write(metrics.get("phase1Snps", "") + '\t')
f.write(metrics.get("phase1SnpsPreserved", "") + '\t')
f.write(metrics.get("snps", "") + '\t')
f.write(metrics.get("snpsPreserved", "") + '\t')
f.write(metrics.get("missingPos", "") + '\t')
f.write(metrics.get("missingPosPreserved", "") + '\t')
f.write(metrics.get("excludedSample", "") + '\t')
f.write(metrics.get("excludedSamplePreserved", "") + '\t')
f.write(quoted(metrics.get("errorList", "")) + '\n')
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)
def collect_metrics(args):
"""Collect the quality metrics and SNP metrics for a sample.
This function expects, or creates '(*)', the following files arranged
in the following way:
reference
referenceFile.fasta
samples
sample_name_one/*.fastq.gz
sample_name_one/reads.sam
sample_name_one/reads.sorted.deduped.bam
sample_name_one/reads.sorted.bam
sample_name_one/reads.all.pileup
sample_name_one/var.flt.vcf
sample_name_one/var.flt_preserved.vcf
sample_name_one/consensus.fasta
sample_name_one/consensus_preserved.fasta
sample_name_one/consensus.vcf
sample_name_one/consensus_preserved.vcf
sample_name_one/metrics*
The input files are created outside of this function. The package
documentation provides an example of preparing these files based on the
lambda_virus sequence that is used as one test for this package.
Parameters
----------
args : argparse.Namespace
referenceFile : File path of the reference fasta file
sampleDir : Relative or absolute directory of the sample
consensusFastaFileName : File name of the consensus fasta file which must exist in the sample directory
consensusPreservedFastaFileName : File name of the consensus preserved fasta file which must exist in the sample directory
consensusVcfFileName : File name of the consensus vcf file which must exist in the sample directory
consensusPreservedVcfFileName : File name of the consensus preserved vcf file which must exist in the sample directory
maxSnps : Maximum allowed number of SNPs per sample
metricsFile : Output file. Relative or absolute path to the metrics file
"""
utils.print_log_header(classpath=True)
utils.print_arguments(args)
#==========================================================================
# Validate inputs
#==========================================================================
# Verify reference fasta file exists and is not empty
reference_file_path = args.referenceFile
utils.verify_non_empty_input_files("Reference file", [reference_file_path],
error_handler="global")
sample_dir = args.sampleDir
utils.verify_non_empty_directory("Sample directory",
sample_dir,
error_handler="sample",
continue_possible=False)
metrics_file_path = args.metricsFile
max_allowed_snps = args.maxSnps
consensus_vcf_file_name = args.consensusVcfFileName
consensus_preserved_vcf_file_name = args.consensusPreservedVcfFileName
consensus_fasta_file_name = args.consensusFastaFileName
consensus_preserved_fasta_file_name = args.consensusPreservedFastaFileName
sample_id = utils.sample_id_from_dir(sample_dir)
#==========================================================================
# Read existing metrics file so some metrics can be reused
#==========================================================================
try:
metrics = utils.read_properties(metrics_file_path)
except IOError:
metrics = dict()
#-------------------------
verbose_print(
"# %s %s" %
(utils.timestamp(), "Get machine and flowcell from fastq header"))
#-------------------------
machine = ""
flowcell = ""
fastq_files = fastq.list_fastq_files(sample_dir)
fastq_files = [f for f in fastq_files
if os.path.isfile(f)] # Exclude broken symlinks
if not fastq_files:
handle_error("No fastq files were found.")
else:
tags = fastq.extract_metadata_tags(fastq_files[0])
if tags:
machine = tags.instrument or ""
flowcell = tags.flow_cell or ""
#-------------------------
verbose_print("# %s %s" %
(utils.timestamp(), "Sum file sizes of paired fastq files"))
#-------------------------
fastq_file_size = ""
fastq_file_list = ""
if fastq_files:
fastq_file_size = sum([os.path.getsize(file) for file in fastq_files])
# Make a comma separated list of just the fastq file names without directories
fastq_file_list = [os.path.basename(file) for file in fastq_files]
fastq_file_list = ", ".join(fastq_file_list)
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(
), "Calculate number of reads, %mapped, %proper pair, and ave insert size from sam file"
))
#-------------------------
num_reads = ""
percent_reads_mapped = ""
percent_proper_pair = ""
ave_insert_size = ""
file = os.path.join(sample_dir, "reads.sam")
if verify_input_file("SAM file", file):
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
num_reads = metrics.get("numberReads",
"") # reuse already fresh metrics
percent_reads_mapped = metrics.get(
"percentReadsMapped", "") # reuse already fresh metrics
percent_proper_pair = metrics.get(
"percentProperPair", "") # reuse already fresh metrics
ave_insert_size = metrics.get("aveInsertSize",
"") # reuse already fresh metrics
missing_any_metrics = not all([
num_reads, percent_reads_mapped, percent_proper_pair,
ave_insert_size
])
if not missing_any_metrics:
verbose_print(
"Reusing previously calculated number of reads, %mapped, %proper pair, and ave insert size"
)
else:
tempfile_path = os.path.join(sample_dir, "tmp.sam.stats")
try:
command.run("samtools stats " + file, tempfile_path)
except subprocess.CalledProcessError:
pass # the error message has already been printed to stderr
with open(tempfile_path) as f:
for line in f:
lower_line = line.lower()
split_line = line.strip().split('\t')
if "raw total sequences:" in lower_line:
num_reads = split_line[2]
continue
if "reads mapped:" in lower_line:
reads_mapped = split_line[2]
try:
percent_reads_mapped = 100.0 * float(
reads_mapped) / float(num_reads)
percent_reads_mapped = "%.2f" % percent_reads_mapped
except ValueError:
percent_reads_mapped = ""
continue
if "reads properly paired:" in lower_line:
proper_pairs = split_line[2]
try:
percent_proper_pair = 100.0 * float(
proper_pairs) / float(num_reads)
percent_proper_pair = "%.2f" % percent_proper_pair
except ValueError:
percent_proper_pair = ""
continue
if "insert size average:" in lower_line:
ave_insert_size = split_line[2]
continue
os.unlink(tempfile_path)
missing_any_metrics = not all([
num_reads, percent_reads_mapped, percent_proper_pair,
ave_insert_size
])
if missing_any_metrics:
missing_list = []
if not num_reads:
missing_list.append("number of reads")
if not percent_reads_mapped:
missing_list.append("percent reads mapped")
if not percent_proper_pair:
missing_list.append("percent proper pair")
if not ave_insert_size:
missing_list.append("ave insert size")
error_text = "Cannot calculate " + ", ".join(
missing_list) + '.'
handle_error(error_text)
#-------------------------
# Calculate number of duplicate reads from deduped bam file
#-------------------------
num_dup_reads = ""
remove_duplicate_reads = os.environ.get("RemoveDuplicateReads") or "true"
remove_duplicate_reads = remove_duplicate_reads.lower()
if remove_duplicate_reads == "true":
verbose_print(
"# %s %s" %
(utils.timestamp(),
"Calculate number of duplicate reads from deduped bam file"))
file = os.path.join(sample_dir, "reads.sorted.deduped.bam")
if verify_input_file("Deduped BAM file", file):
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file],
metrics_file_path)
if not args.forceFlag and not needs_rebuild:
num_dup_reads = metrics.get("numberDupReads",
"") # reuse already fresh metrics
if num_dup_reads:
verbose_print(
"Reusing previously calculated number of duplicate reads")
else:
num_dup_reads = command.run("samtools view -S -c -f 1024 " +
file)
num_dup_reads = num_dup_reads.strip()
#-------------------------
verbose_print("# %s %s" %
(utils.timestamp(), "Calculate mean depth from pileup file"))
#-------------------------
ave_pileup_depth = ""
file = os.path.join(sample_dir, "reads.all.pileup")
if verify_input_file("Pileup file", file):
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
ave_pileup_depth = metrics.get("avePileupDepth",
"") # reuse already fresh metrics
if ave_pileup_depth:
verbose_print("Reusing previously calculated mean pileup depth")
else:
depth_sum = 0
with open(file) as f:
for line in f:
tokens = line.split()
try:
depth_sum += int(tokens[3])
except (ValueError, IndexError):
pass
reference_length = 0
for record in SeqIO.parse(reference_file_path, "fasta"):
reference_length += len(record)
if depth_sum > 0 and reference_length > 0:
#print("depth_sum=%i" % depth_sum);
#print("reference_length=%i" % reference_length)
ave_pileup_depth = float(depth_sum) / float(reference_length)
ave_pileup_depth = "%.2f" % ave_pileup_depth
else:
handle_error("Cannot calculate mean pileup depth.")
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(
), "Count number of high confidence SNP positions from phase 1 vcf file"))
#-------------------------
phase1_snps = ""
excluded_sample = ""
file = os.path.join(sample_dir, "var.flt.vcf")
if verify_input_file("VCF file", file):
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
phase1_snps = metrics.get("phase1Snps",
"") # reuse already fresh metrics
if phase1_snps:
verbose_print("Reusing previously calculated phase1 snps")
else:
phase1_snps = count_vcf_file_snps(file)
# Flag excessive snps
if max_allowed_snps > 0 and phase1_snps > max_allowed_snps:
excluded_sample = "Excluded"
handle_error("Excluded: exceeded %i maxsnps." % max_allowed_snps)
phase1_snps = str(phase1_snps)
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(
), "Count number of filter_regions preserved high confidence SNP positions from phase 1 vcf file"
))
#-------------------------
phase1_snps_preserved = ""
excluded_sample_preserved = ""
file = os.path.join(sample_dir, "var.flt_preserved.vcf")
if verify_input_file("VCF file", file):
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file], metrics_file_path)
if not args.forceFlag and not needs_rebuild:
phase1_snps_preserved = metrics.get(
"phase1SnpsPreserved", "") # reuse already fresh metrics
if phase1_snps_preserved:
verbose_print(
"Reusing previously calculated preserved phase1 snps")
else:
phase1_snps_preserved = count_vcf_file_snps(file)
# Flag excessive snps
if max_allowed_snps > 0 and phase1_snps_preserved > max_allowed_snps:
excluded_sample_preserved = "Excluded"
handle_error("Excluded: preserved exceeded %i maxsnps." %
max_allowed_snps)
phase1_snps_preserved = str(phase1_snps_preserved)
#-------------------------
verbose_print("# %s %s" %
(utils.timestamp(),
"Count number of consensus snps from consensus vcf file"))
#-------------------------
phase2_snps = ""
file = os.path.join(sample_dir, consensus_vcf_file_name)
if verify_input_file("Consensus VCF file", file):
# Omit the phase2 snp count if the sample is excluded.
# It will be meaningless since this sample's phase1 snps are excluded from the snplist.
if excluded_sample != "Excluded":
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file],
metrics_file_path)
if not args.forceFlag and not needs_rebuild:
phase2_snps = metrics.get("snps",
"") # reuse already fresh metrics
if phase2_snps:
verbose_print("Reusing previously calculated phase2 snps")
else:
phase2_snps = count_vcf_file_snps(file)
phase2_snps = str(phase2_snps)
#-------------------------
verbose_print(
"# %s %s" %
(utils.timestamp(),
"Count number of preserved consensus snps from consensus vcf file"))
#-------------------------
phase2_snps_preserved = ""
file = os.path.join(sample_dir, consensus_preserved_vcf_file_name)
if verify_input_file("Consensus VCF file", file):
# Omit the phase2 snp count if the sample is excluded.
# It will be meaningless since this sample's phase1 snps are excluded from the snplist.
if excluded_sample_preserved != "Excluded":
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file],
metrics_file_path)
if not args.forceFlag and not needs_rebuild:
phase2_snps_preserved = metrics.get(
"snpsPreserved", "") # reuse already fresh metrics
if phase2_snps_preserved:
verbose_print(
"Reusing previously calculated preserved phase2 snps")
else:
phase2_snps_preserved = count_vcf_file_snps(file)
phase2_snps_preserved = str(phase2_snps_preserved)
#------------------------------------------
verbose_print(
"# %s %s" %
(utils.timestamp(), "Count missing positions in the snp matrix"))
#------------------------------------------
missing_pos = ""
file = os.path.join(sample_dir, consensus_fasta_file_name)
if verify_input_file("Consensus fasta file", file):
# Omit the phase2 gap count if the sample is excluded.
# It will be meaningless since this sample's phase1 snps are excluded from the snplist.
if excluded_sample != "Excluded":
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file],
metrics_file_path)
if not args.forceFlag and not needs_rebuild:
missing_pos = metrics.get("missingPos",
"") # reuse already fresh metrics
if missing_pos:
verbose_print(
"Reusing previously calculated missing positions")
else:
missing_pos = count_missing_snp_matrix_positions(
file, sample_id)
missing_pos = str(missing_pos)
#------------------------------------------
verbose_print("# %s %s" %
(utils.timestamp(),
"Count missing positions in the preserved snp matrix"))
#------------------------------------------
missing_pos_preserved = ""
file = os.path.join(sample_dir, consensus_preserved_fasta_file_name)
if verify_input_file("Consensus fasta file", file):
# Omit the phase2 gap count if the sample is excluded.
# It will be meaningless since this sample's phase1 snps are excluded from the snplist.
if excluded_sample_preserved != "Excluded":
# Metrics already freshly collected?
needs_rebuild = utils.target_needs_rebuild([file],
metrics_file_path)
if not args.forceFlag and not needs_rebuild:
missing_pos_preserved = metrics.get(
"missingPosPreserved", "") # reuse already fresh metrics
if missing_pos_preserved:
verbose_print(
"Reusing previously calculated missing positions")
else:
missing_pos_preserved = count_missing_snp_matrix_positions(
file, sample_id)
missing_pos_preserved = str(missing_pos_preserved)
#-------------------------
verbose_print("# %s %s" % (utils.timestamp(), "Print results"))
#-------------------------
with open(metrics_file_path, "w") as f:
print("sample=" + '"' + sample_id + '"', file=f)
print("fastqFileList=" + '"' + fastq_file_list + '"', file=f)
print("fastqFileSize=" + str(fastq_file_size), file=f)
print("machine=" + machine, file=f)
print("flowcell=" + flowcell, file=f)
print("numberReads=" + num_reads, file=f)
print("numberDupReads=" + num_dup_reads, file=f)
print("percentReadsMapped=" + percent_reads_mapped, file=f)
print("percentProperPair=" + percent_proper_pair, file=f)
print("aveInsertSize=" + ave_insert_size, file=f)
print("avePileupDepth=" + ave_pileup_depth, file=f)
print("phase1Snps=" + phase1_snps, file=f)
print("phase1SnpsPreserved=" + phase1_snps_preserved, file=f)
print("snps=" + phase2_snps, file=f)
print("snpsPreserved=" + phase2_snps_preserved, file=f)
print("missingPos=" + missing_pos, file=f)
print("missingPosPreserved=" + missing_pos_preserved, file=f)
print("excludedSample=" + excluded_sample, file=f)
print("excludedSamplePreserved=" + excluded_sample_preserved, file=f)
print("errorList=" + '"' + ' '.join(error_list) + '"', file=f)
def combine_metrics(args):
"""Combine the per-sample metrics files into a single table of metrics for all samples.
This function expects, or creates '(*)', the following files arranged
in the following way:
samples
sample_name_one/metrics
metrics.tsv
All the input files are created outside of this function. Before
running this command, the metrics file for each sample must be created by the
collect_metrics command.
The package documentation provides an example of preparing these files based
on the lambda_virus sequence that is used as one test for this package.
Parameters
----------
args : argparse.Namespace
sampleDirsFile : Path to file containing a list of directories -- one per sample
metricsFileName : File name of the metrics files which must exist in each of the sample directories
mergedMetricsFile : Path to the output merged metrics file
"""
utils.print_log_header()
utils.print_arguments(args)
#==========================================================================
# Validate inputs
#==========================================================================
sample_directories_list_path = args.sampleDirsFile
utils.verify_non_empty_input_files("File of sample directories",
[sample_directories_list_path],
error_handler="global")
metrics_file_name = args.metricsFileName
merged_metrics_path = args.mergedMetricsFile
with open(sample_directories_list_path, "r") as f:
sample_directories = [line.rstrip() for line in f]
sample_directories = [d for d in sample_directories if d]
metrics_files = [
os.path.join(d, metrics_file_name) for d in sample_directories
]
#==========================================================================
# Check if merge has already been done
#==========================================================================
needs_rebuild = utils.target_needs_rebuild(metrics_files,
merged_metrics_path)
if not args.forceFlag and not needs_rebuild:
verbose_print(
"# The merged metrics file is already freshly created. Use the -f option to force a rebuild."
)
return
#==========================================================================
# Parse the metrics files and print the tabular results
#==========================================================================
with open(merged_metrics_path, 'w') as f:
# Emit the column headings
column_headings = [
"Sample", "Fastq Files", "Fastq File Size", "Machine", "Flowcell",
"Number of Reads", "Duplicate Reads", "Percent of Reads Mapped",
"Percent Proper Pair", "Average Insert Size",
"Average Pileup Depth", "Phase1 SNPs", "Phase1 Preserved SNPs",
"Phase2 SNPs", "Phase2 Preserved SNPs",
"Missing SNP Matrix Positions",
"Missing Preserved SNP Matrix Positions", "Excluded Sample",
"Excluded Preserved Sample", "Warnings and Errors"
]
if not args.spaceHeadings:
column_headings = [
heading.replace(' ', '_') for heading in column_headings
]
tabbed_headings = '\t'.join(column_headings)
f.write(tabbed_headings + '\n')
# Reads the metrics from each sample, and emit the values
for metrics_file in metrics_files:
verbose_print("Processing " + metrics_file)
message = None
if not os.path.isfile(metrics_file):
message = "Sample metrics file %s does not exist." % metrics_file
elif os.path.getsize(metrics_file) == 0:
message = "Sample metrics file %s is empty." % metrics_file
if message:
f.write(message + '\n')
utils.sample_warning(message)
continue
metrics = utils.read_properties(metrics_file)
f.write(quoted(metrics.get("sample", "")) + '\t')
f.write(quoted(metrics.get("fastqFileList", "")) + '\t')
f.write(metrics.get("fastqFileSize", "") + '\t')
f.write(metrics.get("machine", "") + '\t')
f.write(metrics.get("flowcell", "") + '\t')
f.write(metrics.get("numberReads", "") + '\t')
f.write(metrics.get("numberDupReads", "") + '\t')
f.write(metrics.get("percentReadsMapped", "") + '\t')
f.write(metrics.get("percentProperPair", "") + '\t')
f.write(metrics.get("aveInsertSize", "") + '\t')
f.write(metrics.get("avePileupDepth", "") + '\t')
f.write(metrics.get("phase1Snps", "") + '\t')
f.write(metrics.get("phase1SnpsPreserved", "") + '\t')
f.write(metrics.get("snps", "") + '\t')
f.write(metrics.get("snpsPreserved", "") + '\t')
f.write(metrics.get("missingPos", "") + '\t')
f.write(metrics.get("missingPosPreserved", "") + '\t')
f.write(metrics.get("excludedSample", "") + '\t')
f.write(metrics.get("excludedSamplePreserved", "") + '\t')
f.write(quoted(metrics.get("errorList", "")) + '\n')