def datasets(projects_set): log = task.logger conf = task.conf classifier, projects = projects_set classifier_id = classifier["id"] group_values = classifier["group_values"] short_values = classifier["group_short_values"] long_values = classifier["group_long_values"] group_name = classifier["group_name"] group_short_name = classifier["group_short_name"] group_long_name = classifier["group_long_name"] group_file_prefix = normalize_id(classifier_id) log.info("--- [{0} ({1}) ({2}) ({3})] {4}".format(classifier["name"], group_long_name, group_short_name, group_name, "-" * 30)) log.info("Reading number of samples per project ...") project_ids = [] total_samples = 0 for project in projects: project_id = project["id"] project_ids += [project_id] log.info(" Project {0}".format(project["id"])) projdb = ProjectDb(project["db"]) num_samples = projdb.get_total_affected_samples() total_samples += num_samples log.debug(" {0} samples".format(num_samples)) projdb.close() log.debug(" {0} samples in total".format(total_samples)) log.info("Updating ...") combination_path = get_combination_path(conf) path = os.path.join(combination_path, "{0}.tsv".format(group_file_prefix)) if not os.path.exists(path): with open(path, "w") as f: tsv.write_line(f, "NAME", "SHORT_NAME", "LONG_NAME", "SAMPLES_TOTAL", "PROJECT_IDS") with open(path, "a") as f: tsv.write_line(f, group_name, group_short_name, group_long_name, total_samples, ",".join(project_ids))
def combination_recurrences(projects_set): log = task.logger conf = task.conf classifier, projects = projects_set classifier_id = classifier["id"] group_values = classifier["group_values"] short_values = classifier["group_short_values"] long_values = classifier["group_long_values"] group_name = classifier["group_name"] group_short_name = classifier["group_short_name"] group_long_name = classifier["group_long_name"] if len(group_values) == 0: group_file_prefix = classifier_id else: group_file_prefix = "{0}-{1}".format(classifier_id, group_short_name) group_file_prefix = normalize_id(group_file_prefix) log.info("--- [{0} ({1}) ({2}) ({3})] {4}".format(classifier["name"], group_long_name, group_short_name, group_name, "-" * 30)) log.info("Creating database ...") db_path = make_temp_file(task, suffix="-{0}.db".format(group_file_prefix)) log.debug(" > {0}".format(db_path)) conn = sqlite3.connect(db_path) conn.row_factory = sqlite3.Row create_db(conn) log.info("Combining recurrences ...") c = conn.cursor() sample_total = 0 project_ids = [] for project in projects: project_ids += [project["id"]] log.info(" Project {0}:".format(project["id"])) projdb = ProjectDb(project["db"]) project_sample_total = projdb.get_total_affected_samples() sample_total += project_sample_total log.info(" Total samples = {0}".format(project_sample_total)) log.info(" Variant genes ...") count = 0 for afg in projdb.affected_genes(join_variant=True, join_xrefs=True, join_rec=True): var = afg.var rec = afg.rec if rec.sample_freq is None: log.warn("Discarding variant gene without sample frequency: {0}".format(repr(afg))) continue start, end, ref, alt = var_to_tab(var) try: c.execute("INSERT INTO variants (chr, strand, start, ref, alt, xrefs) VALUES (?,?,?,?,?,?)", (var.chr, var.strand, start, ref, alt, ",".join(var.xrefs))) var_id = c.lastrowid except sqlite3.IntegrityError: c.execute("SELECT var_id FROM variants WHERE chr=? AND strand=? AND start=? AND ref=? AND alt=?", (var.chr, var.strand, start, ref, alt)) r = c.fetchone() var_id = r[0] try: c.execute("INSERT INTO variant_genes (var_id, gene_id, impact, coding_region, prot_changes, sample_freq) VALUES (?,?,?,?,?,?)", (var_id, afg.gene_id, afg.impact, afg.coding_region, afg.prot_changes, rec.sample_freq)) except sqlite3.IntegrityError: c.execute(""" UPDATE variant_genes SET sample_freq=sample_freq + ? WHERE var_id=? AND gene_id=?""", (rec.sample_freq, var_id, afg.gene_id)) count += 1 log.info(" {0} variant genes".format(count)) log.info(" Genes ...") count = 0 for gene in projdb.genes(join_xrefs=True, join_rec=True): rec = gene.rec if rec.sample_freq is None: continue c.execute("SELECT COUNT(*) FROM genes WHERE gene_id=?", (gene.id,)) r = c.fetchone() if r[0] == 0: c.execute("INSERT INTO genes (gene_id, sample_freq) VALUES (?,?)", (gene.id, rec.sample_freq)) else: c.execute("UPDATE genes SET sample_freq=sample_freq + ? WHERE gene_id=?", (rec.sample_freq, gene.id)) count += 1 log.info(" {0} genes".format(count)) log.info(" Pathways ...") count = 0 for pathway in projdb.pathways(join_rec=True): rec = pathway.rec if rec.sample_freq is None: continue c.execute("SELECT COUNT(*) FROM pathways WHERE pathway_id=?", (pathway.id,)) r = c.fetchone() if r[0] == 0: c.execute("INSERT INTO pathways (pathway_id, sample_freq) VALUES (?,?)", (pathway.id, rec.sample_freq)) else: c.execute("UPDATE pathways SET sample_freq=sample_freq + ? WHERE pathway_id=?", (rec.sample_freq, pathway.id)) count += 1 log.info(" {0} pathways".format(count)) projdb.close() log.info("Calculating proportions with {0} samples in total among projects ...".format(sample_total)) if sample_total > 0: c.execute("UPDATE variant_genes SET sample_prop=CAST(sample_freq AS REAL)/{0}.0".format(sample_total)) c.execute("UPDATE genes SET sample_prop=CAST(sample_freq AS REAL)/{0}.0".format(sample_total)) c.execute("UPDATE pathways SET sample_prop=CAST(sample_freq AS REAL)/{0}.0".format(sample_total)) c.close() conn.commit() log.info("Saving results ...") c = conn.cursor() base_path = get_combination_path(conf, "recurrences") log.info(" Variant genes ...") with tsv.open(os.path.join(base_path, "variant_gene-{0}.tsv.gz".format(group_file_prefix)), "w") as f: tsv.write_param(f, "classifier", classifier["id"]) tsv.write_param(f, "group_id", group_name) tsv.write_param(f, "group_short_name", group_short_name) tsv.write_param(f, "group_long_name", group_long_name) tsv.write_param(f, "projects", ",".join(project_ids)) tsv.write_param(f, "SAMPLE_TOTAL", sample_total) tsv.write_line(f, "CHR", "STRAND", "START", "ALLELE", "GENE_ID", "IMPACT", "IMPACT_CLASS", "SAMPLE_FREQ", "SAMPLE_PROP", "PROT_CHANGES", "XREFS") for r in c.execute("SELECT * FROM variant_genes JOIN variants USING (var_id) ORDER BY chr*1, chr, strand, start, gene_id"): strand, ref, alt = r["strand"], r["ref"], r["alt"] allele = "{0}/{1}".format(ref, alt) tsv.write_line(f, r["chr"], strand, r["start"], allele, r["gene_id"], r["impact"], TransFIC.class_name(r["impact"]), r["sample_freq"], r["sample_prop"], r["prot_changes"], r["xrefs"], null_value="-") log.info(" Genes ...") with tsv.open(os.path.join(base_path, "gene-{0}.tsv.gz".format(group_file_prefix)), "w") as f: tsv.write_param(f, "classifier", classifier["id"]) tsv.write_param(f, "group_id", group_name) tsv.write_param(f, "group_short_name", group_short_name) tsv.write_param(f, "group_long_name", group_long_name) tsv.write_param(f, "projects", ",".join(project_ids)) tsv.write_param(f, "SAMPLE_TOTAL", sample_total) tsv.write_line(f, "GENE_ID", "SAMPLE_FREQ", "SAMPLE_PROP") for r in c.execute("SELECT * FROM genes ORDER BY gene_id"): tsv.write_line(f, r["gene_id"], r["sample_freq"], r["sample_prop"], null_value="-") log.info(" Pathways ...") with tsv.open(os.path.join(base_path, "pathway-{0}.tsv.gz".format(group_file_prefix)), "w") as f: tsv.write_param(f, "classifier", classifier["id"]) tsv.write_param(f, "group_id", group_name) tsv.write_param(f, "group_short_name", group_short_name) tsv.write_param(f, "group_long_name", group_long_name) tsv.write_param(f, "projects", ",".join(project_ids)) tsv.write_param(f, "SAMPLE_TOTAL", sample_total) tsv.write_line(f, "PATHWAY_ID", "SAMPLE_FREQ", "SAMPLE_PROP") for r in c.execute("SELECT * FROM pathways ORDER BY pathway_id"): tsv.write_line(f, r["pathway_id"], r["sample_freq"], r["sample_prop"], null_value="-") conn.close() remove_temp(task, db_path)
def combination_oncodrivefm(projects_set): log = task.logger conf = task.conf classifier, projects = projects_set classifier_id = classifier["id"] group_values = classifier["group_values"] short_values = classifier["group_short_values"] long_values = classifier["group_long_values"] group_name = classifier["group_name"] group_short_name = classifier["group_short_name"] group_long_name = classifier["group_long_name"] if len(group_values) == 0: group_file_prefix = classifier_id else: group_file_prefix = "{0}-{1}".format(classifier_id, group_short_name) group_file_prefix = normalize_id(group_file_prefix) log.info("--- [{0} ({1}) ({2}) ({3})] {4}".format(classifier["name"], group_long_name, group_short_name, group_name, "-" * 30)) log.info("Exporting project data ...") base_path = make_temp_dir(task, suffix=".{0}".format(group_file_prefix)) log.debug("> {0}".format(base_path)) project_ids = [] gene_files = [] pathway_files = [] for project in projects: project_id = project["id"] project_ids += [project_id] log.info(" Project {0}:".format(project["id"])) projdb = ProjectDb(project["db"]) log.info(" Genes ...") count = 0 file_path = os.path.join(base_path, "{0}-genes.tsv".format(project_id)) gene_files += [file_path] with open(file_path, "w") as f: tsv.write_param(f, "classifier", classifier_id) tsv.write_param(f, "group_id", group_name) tsv.write_param(f, "slice", project_id) tsv.write_line(f, "GENE_ID", "PVALUE") for gene in projdb.genes(): if gene.fm_pvalue is not None: tsv.write_line(f, gene.id, gene.fm_pvalue, null_value="-") count += 1 log.info(" {0} genes".format(count)) log.info(" Pathways ...") count = 0 file_path = os.path.join(base_path, "{0}-pathways.tsv".format(project_id)) pathway_files += [file_path] with open(file_path, "w") as f: tsv.write_param(f, "classifier", classifier_id) tsv.write_param(f, "group_id", group_name) tsv.write_param(f, "slice", project_id) tsv.write_line(f, "PATHWAY_ID", "ZSCORE") for pathway in projdb.pathways(): if pathway.fm_zscore is not None: tsv.write_line(f, pathway.id, pathway.fm_zscore, null_value="-") count += 1 log.info(" {0} pathways".format(count)) projdb.close() log.info("Combining ...") combination_path = get_combination_path(conf, "oncodrivefm") log.info(" Genes ...") cmd = " ".join([ "oncodrivefm-combine", "-m median-empirical", "-o '{0}'".format(combination_path), "-n 'gene-{0}'".format(group_file_prefix), "-D 'classifier={0}'".format(classifier_id), "-D 'group_id={0}'".format(group_name), "-D 'group_short_name={0}'".format(group_short_name), "-D 'group_long_name={0}'".format(group_long_name), "--output-format tsv.gz" ] + ["'{0}'".format(name) for name in gene_files]) log.debug(cmd) ret_code = subprocess.call(cmd, shell=True) if ret_code != 0: #log.error("OncodriveFM error while combining gene pvalues:\n{0}".format(cmd)) #return -1 raise Exception("OncodriveFM error while combining gene pvalues:\n{0}".format(cmd)) log.info(" Pathways ...") cmd = " ".join([ "oncodrivefm-combine", "-m median-zscore", "-o '{0}'".format(combination_path), "-n 'pathway-{0}'".format(group_file_prefix), "-D 'classifier={0}'".format(classifier_id), "-D 'group_id={0}'".format(group_name), "-D 'group_short_name={0}'".format(group_short_name), "-D 'group_long_name={0}'".format(group_long_name), "--output-format tsv.gz" ] + ["'{0}'".format(name) for name in pathway_files]) log.debug(cmd) ret_code = subprocess.call(cmd, shell=True) if ret_code != 0: #log.error("OncodriveFM error while combining pathway zscores:\n{0}".format(cmd)) #return -1 raise Exception("OncodriveFM error while combining pathway zscores:\n{0}".format(cmd)) remove_temp(task, base_path)
def drivers(): log = task.logger conf = task.conf db_path = get_results_path(conf, "drivers.db") db = SigDb(db_path) db.open() log.info("Variants ...") path = get_combination_path(conf, "recurrences", "variant_gene-global-all.tsv.gz") with tsv.open(path, "r") as f: types = (str, str, int, str) for fields in tsv.lines(f, types, columns=("CHR", "STRAND", "START", "ALLELE"), header=True): chr, strand, start, allele = fields[:4] db.add_variant(chr, start) log.info("Genes ...") gene_sites = {} gene_fm = set() gene_clust = set() #SPECIAL_THRESHOLD = ["C18", "C34"] SPECIAL_THRESHOLD = [] log.info(" OncodriveFM ...") filename_re = re.compile(r"gene-cancer_site-(.+)\.tsv.gz") base_path = get_combination_path(conf, "oncodrivefm") for path in os.listdir(base_path): m = filename_re.match(path) if not m: continue cancer_site_code = m.group(1) if cancer_site_code in SPECIAL_THRESHOLD: threshold = 1e-6 else: threshold = 0.01 with tsv.open(os.path.join(base_path, path), "r") as f: params = tsv.params(f) cancer_site_name = params["group_long_name"] for fields in tsv.lines(f, (str, float), columns=("ID", "QVALUE"), header=True): gene, qvalue = fields if qvalue < threshold: add_cancer_site(gene_sites, gene, cancer_site_code, cancer_site_name) gene_fm.add(gene) log.info(" OncodriveCLUST ...") filename_re = re.compile(r"cancer_site-(.+)\.tsv.gz") base_path = get_combination_path(conf, "oncodriveclust") for path in os.listdir(base_path): m = filename_re.match(path) if not m: continue cancer_site_code = m.group(1) with tsv.open(os.path.join(base_path, path), "r") as f: params = tsv.params(f) cancer_site_name = params["group_long_name"] for fields in tsv.lines(f, (str, float), columns=("ID", "QVALUE"), header=True): gene, qvalue = fields if qvalue < 0.05: add_cancer_site(gene_sites, gene, cancer_site_code, cancer_site_name) gene_clust.add(gene) log.info(" Updating db ...") sig_genes = gene_fm | gene_clust for gene in sig_genes: db.add_gene(gene, gene in gene_fm, gene in gene_clust) log.info("Saving driver genes cancer sites dataset ...") path = get_results_path(conf, "gene-driver_cancer_sites.tsv") log.debug("> {}".format(path)) with open(path, "w") as f: tsv.write_param(f, "date", datetime.now()) tsv.write_line(f, "GENE_ID", "FM", "CLUST", "CANCER_SITES_COUNT", "CANCER_SITE_CODES", "CANCER_SITE_NAMES") for gene, sites in gene_sites.items(): tsv.write_line(f, gene, 1 if gene in gene_fm else 0, 1 if gene in gene_clust else 0, len(sites), ", ".join(sorted([code for code, name in sites])), ", ".join(sorted([name for code, name in sites]))) db.commit() db.close()