def create_normal_db(coverage_files_txt, snv_pon, out_dir, genome_build): """create normal db input: coverage files calculated by purecn for each sample snv_pon - mutect2 SNV PON output: mapping_bias_hg38.rds normalDB_hg38.rds """ rscript = utils.Rscript_cmd("r36") normaldb_r = utils.R_package_script("r36", "PureCN", "extdata/NormalDB.R") cmd = [rscript, normaldb_r, "--outdir", out_dir, "--coveragefiles", coverage_files_txt, "--normal_panel" , snv_pon, "--genome", genome_build, "--force"] try: cmd_line = "export R_LIBS_USER=%s && %s && %s" % (utils.R_sitelib(env = "r36"), utils.get_R_exports(env = "r36"), " ".join([str(x) for x in cmd])) do.run(cmd_line, "PureCN normalDB") except subprocess.CalledProcessError as msg: logger.info("PureCN failed to create a normal db") return out_dir
def get_coverage(data): """Calculate coverage for a sample.bam, account for GC content data is single sample """ data = utils.to_single_data(data) bed_file = tz.get_in(["config", "algorithm", "purecn_bed_ready"], data) sample_name = dd.get_sample_name(data) work_dir = _sv_workdir(data) rscript = utils.Rscript_cmd("r36") coverage_r = utils.R_package_script("r36", "PureCN", "extdata/Coverage.R") intervals = tz.get_in(["config", "algorithm", "purecn_bed_ready"], data) # PureCN resolves symlinks and the actual output PureCN coverage file name # is derived from the end bam not from bam_file bam_file = os.path.realpath(dd.get_align_bam(data)) bam_name = os.path.basename(bam_file) (bname, ext) = os.path.splitext(bam_name) result_file = os.path.join(work_dir, bname + "_coverage_loess.txt.gz") if not os.path.exists(result_file): cmd = [rscript, coverage_r, "--outdir", work_dir, "--bam", bam_file, "--intervals", intervals] try: cmd_line = "export R_LIBS_USER=%s && %s && %s" % (utils.R_sitelib(env = "r36"), utils.get_R_exports(env = "r36"), " ".join([str(x) for x in cmd])) do.run(cmd_line, "PureCN coverage") except subprocess.CalledProcessError as msg: logger.info("PureCN failed to calculate coverage") logger.debug("Saved PureCN coverage files to " + result_file) return result_file
def process_intervals(data): """Prepare intervals file""" bed_file = regions.get_sv_bed(data) if not bed_file: bed_file = bedutils.clean_file(dd.get_variant_regions(data), data) if not bed_file: return None basename = os.path.splitext(bed_file)[0] ready_file = basename + ".txt" if os.path.exists(ready_file): return ready_file optimized_bed = basename + ".optimized.bed" rscript = utils.Rscript_cmd("r36") interval_file_r = utils.R_package_script("r36", "PureCN", "extdata/IntervalFile.R") ref_file = dd.get_ref_file(data) mappability_resource = dd.get_variation_resources(data)["purecn_mappability"] genome = dd.get_genome_build(data) cmd = [rscript, interval_file_r, "--infile", bed_file, "--fasta", ref_file, "--outfile", ready_file, "--offtarget", "--genome", genome, "--export", optimized_bed, "--mappability", mappability_resource] try: cmd_line = "export R_LIBS_USER=%s && %s && %s" % (utils.R_sitelib(env = "r36"), utils.get_R_exports(env = "r36"), " ".join([str(x) for x in cmd])) do.run(cmd_line, "PureCN intervals") except subprocess.CalledProcessError as msg: logger.info("PureCN failed to prepare intervals") logger.debug("Saved PureCN interval file into " + ready_file) return ready_file
def _run_purecn_dx(out, paired): """Extract signatures and mutational burdens from PureCN rds file.""" # no solution - no signatures if not "rds" in out: return out rscript = utils.Rscript_cmd() purecndx_r = utils.R_package_script("PureCN", "extdata/Dx.R", env="base") simple_repeat_bed = dd.get_variation_resources( paired.tumor_data)["simple_repeat"] callable_bed = dd.get_sample_callable(paired.tumor_data) out_base = utils.splitext_plus(out["rds"])[0] mutation_burden_csv = out_base + "_mutation_burden.csv" if not utils.file_uptodate(mutation_burden_csv, out["rds"]): # no signatures - so we generate them with file_transaction(paired.tumor_data, out_base) as tx_out_base: cmd = [ rscript, purecndx_r, "--rds", out["rds"], "--callable", callable_bed, "--signatures", "--exclude", simple_repeat_bed, "--out", tx_out_base ] do.run(cmd, "PureCN Dx mutational burden and signatures") out_base, out, all_files = _get_purecn_dx_files(paired, out, require_exist=True) # if a file was not generated it would not go to the upload for f in all_files: if os.path.exists(os.path.join(os.path.dirname(tx_out_base), f)): shutil.move(os.path.join(os.path.dirname(tx_out_base), f), os.path.join(os.path.dirname(out_base), f)) return out
def _run_purecn_normaldb(paired, out): """Run PureCN with normaldb and native segmentation paired is one t/n pair or only """ sample = utils.to_single_data(paired.tumor_data) bed_file = tz.get_in(["config", "algorithm", "purecn_bed_ready"], sample) sample_name = dd.get_sample_name(sample) work_dir = _sv_workdir(sample) rscript = utils.Rscript_cmd("r36") purecn_r = utils.R_package_script("r36", "PureCN", "extdata/PureCN.R") intervals = tz.get_in(["config", "algorithm", "purecn_bed_ready"], sample) bam_file = dd.get_align_bam(sample) # termline and somatic - just annotated and filters assigned variants_vcf = tz.get_in(["variants"], sample)[0].get("germline") # in a T/N case, there is no germline file - vrn file with all variants if not variants_vcf: variants_vcf = tz.get_in(["variants"], sample)[0].get("vrn_file") normaldb = tz.get_in(["config", "algorithm", "background", "cnv_reference", "purecn_normaldb"], sample) mappingbiasfile = tz.get_in(["config", "algorithm", "background", "cnv_reference", "purecn_mapping_bias"], sample) sample_coverage = tz.get_in(["depth", "bins", "purecn"], sample) simple_repeat_bed = dd.get_variation_resources(sample)["simple_repeat"] result_file = os.path.join(work_dir, sample_name + ".rds") genome = dd.get_genome_build(sample) cmd = [ rscript, purecn_r, "--out", work_dir, "--tumor", sample_coverage, "--sampleid", sample_name, "--vcf", variants_vcf, "--normaldb", normaldb, "--mappingbiasfile", mappingbiasfile, "--intervals", intervals, "--snpblacklist", simple_repeat_bed, "--genome", genome, "--force", "--postoptimize", "--seed", "123", "--bootstrapn", "500", "--cores", dd.get_num_cores(sample)] resources = config_utils.get_resources("purecn", sample) if "options" in resources: cmd += [str(x) for x in resources.get("options", [])] # it is not recommended to use matched normal sample in PureCN analysis, # because then it skips PON coverage normalization and denoising steps! # but still, if it is supplied, we useit if paired.normal_data: normal_sample = utils.to_single_data(paired.normal_data) if normal_sample: normal_coverage = tz.get_in(["depth", "bins", "purecn"], normal_sample) cmd.extend(["--normal", normal_coverage]) if not os.path.exists(result_file): try: cmd_line = "export R_LIBS_USER=%s && %s && %s" % (utils.R_sitelib(env = "r36"), utils.get_R_exports(env = "r36"), " ".join([str(x) for x in cmd])) do.run(cmd_line, "PureCN copy number calling") logger.debug("Saved PureCN output to " + work_dir) except subprocess.CalledProcessError as msg: logger.info("PureCN failed") out_base, out, all_files = _get_purecn_files(paired, work_dir, require_exist = True) return out
def _run_purecn_dx(out, paired): """Extract signatures and mutational burdens from PureCN rds file.""" out_base, out, all_files = _get_purecn_dx_files(paired, out) rscript = utils.Rscript_cmd("r36") purecndx_r = utils.R_package_script("r36", "PureCN", "extdata/Dx.R") simple_repeat_bed = dd.get_variation_resources(paired.tumor_data)["simple_repeat"] callable_bed = dd.get_sample_callable(paired.tumor_data) if not utils.file_uptodate(out["mutation_burden"], out["rds"]): with file_transaction(paired.tumor_data, out_base) as tx_out_base: cmd = [rscript, purecndx_r, "--rds", out["rds"], "--callable", callable_bed, "--signatures", "--exclude", simple_repeat_bed, "--out", tx_out_base] do.run(cmd, "PureCN Dx mutational burden and signatures") for f in all_files: if os.path.exists(os.path.join(os.path.dirname(tx_out_base), f)): shutil.move(os.path.join(os.path.dirname(tx_out_base), f), os.path.join(os.path.dirname(out_base), f)) return out
def _run_purecn(paired, work_dir): """Run PureCN.R wrapper with pre-segmented CNVkit or GATK4 inputs. """ segfns = { "cnvkit": _segment_normalized_cnvkit, "gatk-cnv": _segment_normalized_gatk } out_base, out, all_files = _get_purecn_files(paired, work_dir) failed_file = out_base + "-failed.log" cnr_file = tz.get_in(["depth", "bins", "normalized"], paired.tumor_data) if not utils.file_uptodate( out["rds"], cnr_file) and not utils.file_exists(failed_file): cnr_file, seg_file = segfns[cnvkit.bin_approach(paired.tumor_data)]( cnr_file, work_dir, paired) from bcbio import heterogeneity vcf_file = heterogeneity.get_variants( paired.tumor_data, include_germline=False)[0]["vrn_file"] vcf_file = germline.filter_to_pass_and_reject(vcf_file, paired, out_dir=work_dir) with file_transaction(paired.tumor_data, out_base) as tx_out_base: # Use UCSC style naming for human builds to support BSgenome genome = ("hg19" if dd.get_genome_build(paired.tumor_data) in [ "GRCh37", "hg19" ] else dd.get_genome_build(paired.tumor_data)) rscript = utils.Rscript_cmd() purecn_r = utils.R_package_script("PureCN", "extdata/PureCN.R", env="base") cmd = [ rscript, purecn_r, "--seed", "42", "--out", tx_out_base, "--rds", "%s.rds" % tx_out_base, "--sampleid", dd.get_sample_name(paired.tumor_data), "--genome", genome, "--vcf", vcf_file, "--tumor", cnr_file, "--segfile", seg_file, "--funsegmentation", "Hclust", "--maxnonclonal", "0.3" ] if dd.get_num_cores(paired.tumor_data) > 1: cmd += ["--cores", str(dd.get_num_cores(paired.tumor_data))] try: cmd = "export R_LIBS_USER=%s && %s && %s" % (utils.R_sitelib( env="base"), utils.get_R_exports(env="base"), " ".join( [str(x) for x in cmd])) do.run(cmd, "PureCN copy number calling") except subprocess.CalledProcessError as msg: if _allowed_errors(str(msg)): logger.info( "PureCN failed to find solution for %s: skipping" % dd.get_sample_name(paired.tumor_data)) with open(failed_file, "w") as out_handle: out_handle.write(str(msg)) else: logger.exception() raise for f in all_files: if os.path.exists(os.path.join(os.path.dirname(tx_out_base), f)): shutil.move(os.path.join(os.path.dirname(tx_out_base), f), os.path.join(os.path.dirname(out_base), f)) out = _get_purecn_files(paired, work_dir, require_exist=True)[1] return out if (out.get("rds") and os.path.exists(out["rds"])) else None