def create(project):
log = task.logger
config = GlobalConfig(task.conf)
paths = PathsConfig(config)
project_id = project["id"]
log.info("--- [{0}] --------------------------------------------".format(project_id))
project_path = project["path"]
temp_path = project["temp_path"]
templates_path = config.website.templates_path
if templates_path is None:
log.warn("No website templates have been defined in the configuration. Skipping website creation.")
return
log.info("Creating website ...")
website_path = paths.project_website_path(project_path)
if os.path.exists(website_path):
shutil.rmtree(website_path)
log.info("Copying templates ...")
shutil.copytree(templates_path, website_path)
gitignore_path = os.path.join(website_path, ".gitignore")
try:
os.remove(gitignore_path)
except:
pass
log.info("Expanding templates ...")
vars = dict(
PROJECT_NAME=project_id,
SHOW_ALL_TABS=not config.variants_only)
tmpl_paths = [
os.path.join(website_path, "css", "header.html"),
os.path.join(website_path, "onexus-project.onx")
]
for path in tmpl_paths:
with open(path, "r") as f:
t = Template(f.read())
with open(path, "w") as f:
f.write(t.safe_substitute(vars))
# Creating a soft link to the results folder
project_results_path = paths.project_results_path(project_path)
os.symlink(project_results_path, os.path.join(website_path, "results"))
# Send project to the next modules
projects_port = task.ports("projects_out")
project["website"] = website_path
projects_port.send(project)
def scan_projects(project_out):
log = task.logger
config = GlobalConfig(task.conf)
paths = PathsConfig(config)
projects_path = paths.projects_path()
log.info("Scanning projects ...")
count = 0
sent_projects = []
for path, project in list_projects(log, projects_path):
project_path = os.path.dirname(path)
if "id" not in project:
log.warn("Discarding project that doesn't have 'id': {0}".format(path))
continue
project_id = project["id"]
if "name" in project:
log.info("--- [{0}: {1}] ---------------------------------".format(project_id, project["name"]))
else:
log.info("--- [{0}] ---------------------------------".format(project_id))
if "db" not in project:
project["db"] = os.path.join(project_path, "project.db.gz")
elif not os.path.isabs(project["db"]):
project["db"] = os.path.join(project_path, project["db"])
if not os.path.exists(project["db"]):
log.error("Project database not found at {0}".format(os.path.relpath(project["db"], project_path)))
continue
if project["db"].endswith(".gz"):
log.info("Uncompressing project database ...")
retcode = subprocess.call("gunzip -fc {0} >{1}".format(project["db"], project["db"][:-3]), shell=True)
if retcode != 0:
log.error("Unexpected error while uncompressing the project database at {0}".format(os.path.relpath(project["db"], projects_path)))
continue
project["db"] = project["db"][:-3]
temp_path = paths.project_temp_path(project_path)
if not os.path.exists(temp_path):
os.makedirs(temp_path)
project["path"] = project_path
project["temp_path"] = temp_path
sent_projects += [project_id]
project_out.send(project)
count += 1
log.info("Found {0} projects:\n - {1}".format(count, "\n - ".join(sent_projects)))
def datasets(projects_set):
log = task.logger
config = GlobalConfig(task.conf)
paths = PathsConfig(config)
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 = paths.combination_path()
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 oncoclust(project):
log = task.logger
config = GlobalConfig(task.conf)
paths = PathsConfig(config)
projects_out_port = task.ports("projects_out")
project_id = project["id"]
log.info("--- [{0}] --------------------------------------------".format(project_id))
gene_transcripts_path = paths.data_ensembl_gene_transcripts_path()
oclust = project["oncodriveclust"]
data_paths = oclust["data_paths"]
samples_threshold = oclust["samples_threshold"]
oclust["results"] = os.path.join(project["temp_path"], "oncodriveclust-results.tsv")
cmd = " ".join([
"oncodriveclust",
"-c",
"-m", str(samples_threshold),
"-o", oclust["results"],
data_paths[0],
data_paths[1],
gene_transcripts_path
])
log.debug(cmd)
ret_code = subprocess.call(cmd, shell=True)
if ret_code == 1:
log.warn("No results were generated")
elif ret_code != 0:
log.error("Error while executing OncodriveCLUST:\n{0}".format(cmd))
projects_out_port.send(project)
def finalize_all():
log = task.logger
config = GlobalConfig(task.conf)
paths = PathsConfig(config)
if config.results.use_storage:
log.info("Uploading combination files ...")
combination_path = paths.combination_path()
exclude = ["sources/*", "project.db", "project.db.gz"]
object_prefix = "results/combination"
start_callback = lambda path: log.info(" {}".format(path))
task.storage.upload(combination_path, object_prefix=object_prefix, exclude=exclude, overwrite=True,
start_callback=start_callback)
if config.results.purge_after_upload:
log.info("Purging combination files ...")
shutil.rmtree(combination_path)
def combination_recurrences(projects_set):
log = task.logger
config = GlobalConfig(task.conf)
paths = PathsConfig(config)
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 = paths.combination_path("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 scan_files(project):
log = task.logger
conf = task.conf
config = GlobalConfig(conf)
paths = PathsConfig(config)
projects_port, liftover_projects_port = task.ports("projects_out", "liftover_projects")
project_id = project["id"]
temp_path = project["temp_path"]
project_path = project["path"]
projdb_path = project["db"]
assembly = project["assembly"]
log.info("--- [{0}] --------------------------------------------".format(project_id))
if assembly == "hg18":
out_port = liftover_projects_port
elif assembly == "hg19":
out_port = projects_port
else:
raise Exception("Unexpected assembly: {0}".format(assembly))
#if os.path.exists(projdb_path):
# log.warn("Variations database already created, skipping this step.")
# out_port.send(project)
# return
if os.path.exists(projdb_path):
os.remove(projdb_path)
log.info("Creating variants database ...")
projdb_tmp_path = make_temp_file(task, suffix=".db")
log.debug(projdb_tmp_path)
projdb = ProjectDb(projdb_tmp_path).create()
data_path = config.data_path
log.info("Loading genes ...")
projdb.load_genes(paths.data_ensembl_genes_path())
log.info("Loading pathways ...")
projdb.load_pathways(
paths.data_kegg_def_path(),
paths.data_kegg_ensg_map_path())
log.info("Parsing variants ...")
for obj_name in project["storage_objects"]:
log.info("Downloading {} ...".format(obj_name))
dst_path = os.path.join(project_path, "sources", os.path.basename(obj_name))
dst_dirname = os.path.dirname(dst_path)
if not os.path.exists(dst_dirname):
os.makedirs(dst_dirname)
# TODO: do not copy the source file (do not specify dst_path)
task.storage.get_object(obj_name).get_data(dst_path)
for container_name, path, name, ext, f in archived_files(dst_path):
fname = os.path.join(path, name + ext)
if container_name is not None:
source_name = "{0}:{1}".format(os.path.basename(container_name), fname)
else:
source_name = name + ext
log.info("=> {0} ...".format(source_name))
sample_id = os.path.basename(name)
if ext.lower() in _SUPPORTED_EXTENSIONS:
parser_type = ext[1:]
else:
parser_type = "tab"
parser = create_variants_parser(parser_type, f, source_name, sample_id)
source_id = projdb.add_source(source_name)
var_ids = set()
for var in parser:
for line_num, text in parser.read_lines():
projdb.add_source_line(source_id, line_num, text)
var_id = projdb.add_variant(var, source_id=source_id, line_num=parser.get_line_num())
var_ids.add(var_id)
for line_num, text in parser.read_lines():
projdb.add_source_line(source_id, line_num, text)
num_variants = len(var_ids)
log.info(" {0} variants".format(num_variants))
if num_variants == 0:
raise Exception("No variants found in source '{}'. "
"Please check the documentation for the expected input for '{}' format.".format(
source_name, parser.name))
projdb.commit()
projdb.close()
log.info("Copying variants database ...")
log.debug("{0} -> {1}".format(projdb_tmp_path, projdb_path))
shutil.copy(projdb_tmp_path, projdb_path)
remove_temp(task, projdb_tmp_path)
out_port.send(project)
def create_datasets(project):
log = task.logger
config = GlobalConfig(task.conf)
paths = PathsConfig(config)
project_id = project["id"]
log.info("--- [{0}] --------------------------------------------".format(project_id))
project_path = project["path"]
temp_path = project["temp_path"]
datasets_path = paths.project_results_path(project_path)
ensure_path_exists(datasets_path)
sigdb = SigDb(config.sigdb_path)
sigdb.open()
projdb = ProjectDb(project["db"])
gene_sym = projdb.get_gene_symbols()
total_samples = projdb.get_total_affected_samples()
log.info("Exporting variant genes ...")
vf = open_dataset(project_id, project_path, datasets_path, "variant_gene", "w", log)
tsv.write_param(vf, "SAMPLE_TOTAL", total_samples)
tsv.write_line(vf, "VAR_ID", "CHR", "STRAND", "START", "ALLELE",
"GENE_ID", "IMPACT", "IMPACT_CLASS",
"SAMPLE_FREQ", "SAMPLE_PROP",
"CODING_REGION", "PROTEIN_CHANGES", "INTOGEN_DRIVER", "XREFS")
sf = open_dataset(project_id, project_path, datasets_path, "variant-samples", "w", log)
tsv.write_line(sf, "VAR_ID", "CHR", "STRAND", "START", "ALLELE", "SAMPLE")
count = 0
for afg in projdb.affected_genes(join_variant=True, join_samples=True, join_xrefs=True, join_rec=True):
var = afg.var
rec = afg.rec
start, end, ref, alt = var_to_tab(var)
allele = "{0}/{1}".format(ref, alt)
xrefs = [xref for xref in var.xrefs]
if sigdb.exists_variant(var.chr, start):
xrefs += ["I:1"]
xrefs = ",".join(xrefs)
intogen_driver = 1 if sigdb.exists_gene(afg.gene_id) else 0
tsv.write_line(vf, var.id, var.chr, var.strand, start, allele,
afg.gene_id, afg.impact, TransFIC.class_name(afg.impact),
rec.sample_freq, rec.sample_prop,
afg.coding_region, afg.prot_changes, intogen_driver, xrefs, null_value="\N")
for sample in var.samples:
tsv.write_line(sf, var.id, var.chr, var.strand, start, allele, sample.name, null_value="\N")
count += 1
vf.close()
sf.close()
log.info(" {0} variant genes".format(count))
log.info("Exporting consequences ...")
cf = open_dataset(project_id, project_path, datasets_path, "consequence", "w", log)
tsv.write_line(cf, "VAR_ID", "CHR", "STRAND", "START", "ALLELE", "TRANSCRIPT_ID",
"CT", "GENE_ID", "SYMBOL", "UNIPROT_ID", "PROTEIN_ID", "PROTEIN_POS", "AA_CHANGE",
"SIFT_SCORE", "SIFT_TRANSFIC", "SIFT_TRANSFIC_CLASS",
"PPH2_SCORE", "PPH2_TRANSFIC", "PPH2_TRANSFIC_CLASS",
"MA_SCORE", "MA_TRANSFIC", "MA_TRANSFIC_CLASS",
"IMPACT", "IMPACT_CLASS")
count = 0
for csq in projdb.consequences(join_variant=True):
var = csq.var
start, end, ref, alt = var_to_tab(var)
allele = "{0}/{1}".format(ref, alt)
uniprot = protein = protein_pos = aa_change = None
sift_score = sift_tfic = sift_tfic_class = None
pph2_score = pph2_tfic = pph2_tfic_class = None
ma_score = ma_tfic = ma_tfic_class = None
if so.match(csq.ctypes, so.ONCODRIVEFM):
uniprot, protein = csq.uniprot, csq.protein
if so.match(csq.ctypes, so.NON_SYNONYMOUS):
protein_pos, aa_change = csq.protein_pos, csq.aa_change
sift_score, sift_tfic, sift_tfic_class = csq.sift_score, csq.sift_tfic, TransFIC.class_name(csq.sift_tfic_class)
pph2_score, pph2_tfic, pph2_tfic_class = csq.pph2_score, csq.pph2_tfic, TransFIC.class_name(csq.pph2_tfic_class)
ma_score, ma_tfic, ma_tfic_class = csq.ma_score, csq.ma_tfic, TransFIC.class_name(csq.ma_tfic_class)
tsv.write_line(cf, var.id, var.chr, var.strand, start, allele, csq.transcript,
",".join(csq.ctypes), csq.gene, gene_sym.get(csq.gene),
uniprot, protein, protein_pos, aa_change,
sift_score, sift_tfic, sift_tfic_class,
pph2_score, pph2_tfic, pph2_tfic_class,
ma_score, ma_tfic, ma_tfic_class,
csq.impact, TransFIC.class_name(csq.impact), null_value="\N")
count += 1
cf.close()
log.info(" {0} consequences".format(count))
log.info("Exporting genes ...")
gf = open_dataset(project_id, project_path, datasets_path, "gene", "w", log)
tsv.write_param(gf, "SAMPLE_TOTAL", total_samples)
tsv.write_line(gf, "GENE_ID", "FM_PVALUE", "FM_QVALUE", "FM_EXC_CAUSE",
"CLUST_ZSCORE", "CLUST_PVALUE", "CLUST_QVALUE", "CLUST_EXC_CAUSE", "CLUST_COORDS",
"SAMPLE_FREQ", "SAMPLE_PROP", "INTOGEN_DRIVER")
for gene in projdb.genes(join_rec=True):
rec = gene.rec
if rec.sample_freq is None or rec.sample_freq == 0:
continue
intogen_driver = 1 if sigdb.exists_gene(gene.id) else 0
tsv.write_line(gf, gene.id, gene.fm_pvalue, gene.fm_qvalue, gene.fm_exc_cause,
gene.clust_zscore, gene.clust_pvalue, gene.clust_qvalue, gene.clust_exc_cause, gene.clust_coords,
rec.sample_freq or 0, rec.sample_prop or 0,
intogen_driver, null_value="\N")
gf.close()
log.info("Exporting pathways ...")
pf = open_dataset(project_id, project_path, datasets_path, "pathway", "w", log)
tsv.write_param(pf, "SAMPLE_TOTAL", total_samples)
tsv.write_line(pf, "PATHWAY_ID", "GENE_COUNT", "FM_ZSCORE", "FM_PVALUE", "FM_QVALUE",
"SAMPLE_FREQ", "SAMPLE_PROP", "GENE_FREQ", "GENE_TOTAL", "GENE_PROP")
for pathway in projdb.pathways(join_rec=True):
rec = pathway.rec
if rec.sample_freq is None or rec.sample_freq == 0:
continue
tsv.write_line(pf, pathway.id, pathway.gene_count, pathway.fm_zscore, pathway.fm_pvalue, pathway.fm_qvalue,
rec.sample_freq or 0, rec.sample_prop or 0, rec.gene_freq or 0, pathway.gene_count, rec.gene_prop or 0, null_value="\N")
pf.close()
if not config.skip_oncodrivefm:
log.info("Exporting genes per sample functional impact ...")
with open_dataset(project_id, project_path, datasets_path, "gene_sample-fimpact", "w", log) as f:
tsv.write_line(f, "GENE_ID", "SAMPLE",
"SIFT_SCORE", "SIFT_TRANSFIC", "SIFT_TRANSFIC_CLASS",
"PPH2_SCORE", "PPH2_TRANSFIC", "PPH2_TRANSFIC_CLASS",
"MA_SCORE", "MA_TRANSFIC", "MA_TRANSFIC_CLASS")
for fields in projdb.sample_gene_fimpacts():
(gene, sample,
sift_score, sift_tfic, sift_tfic_class,
pph2_score, pph2_tfic, pph2_tfic_class,
ma_score, ma_tfic, ma_tfic_class) = fields
tsv.write_line(f, gene, sample,
sift_score, sift_tfic, TransFIC.class_name(sift_tfic_class),
pph2_score, pph2_tfic, TransFIC.class_name(pph2_tfic_class),
ma_score, ma_tfic, TransFIC.class_name(ma_tfic_class), null_value="\N")
projdb.close()
sigdb.close()
log.info("Saving project configuration ...")
projres = ProjectResults(project)
with open_dataset(project_id, project_path, datasets_path, "project.tsv", "w", log) as f:
names = ["ASSEMBLY", "SAMPLES_TOTAL"]
values = [project["assembly"], total_samples]
names, values = projres.get_annotations_to_save(config.project.annotations, project["annotations"], names=names, values=values)
tsv.write_line(f, *names)
tsv.write_line(f, *values, null_value="\N")
projects_port = task.ports("projects_out")
projects_port.send(project)
def compute(project):
log = task.logger
config = GlobalConfig(task.conf)
paths = PathsConfig(config)
projects_out_port = task.ports("projects_out")
project_id = project["id"]
log.info("--- [{0}] --------------------------------------------".format(project_id))
ofm = Data.element(project["oncodrivefm"])
feature = ofm["feature"]
slice_name = ofm["slice"]
estimator = ofm.get("estimator")
num_cores = ofm.get("num_cores", dtype=str)
num_samplings = ofm.get("num_samplings", dtype=str)
samples_threshold = ofm.get("samples_threshold", dtype=str)
filter_enabled = ofm.get("filter_enabled", dtype=bool)
filter_path = ofm.get("filter_path", dtype=str)
log.info("feature = {0}".format(feature))
log.info("slice = {0}".format(slice_name))
log.info("estimator = {0}".format(estimator))
log.info("num_cores = {0}".format(num_cores))
log.info("num_samplings = {0}".format(num_samplings))
log.info("samples_threshold = {0}".format(samples_threshold))
log.info("filter_enabled = {0}".format(filter_enabled))
log.info("filter_path = {0}".format(os.path.basename(filter_path)))
cmd = [
"oncodrivefm-compute",
"-o", project["temp_path"],
"-n oncodrivefm-{0}".format(feature),
"-N", num_samplings,
"--threshold", samples_threshold,
"-e {0}".format(estimator),
"-j", num_cores,
"--slices '{0}'".format(slice_name)]
if filter_enabled:
cmd += ["--filter", filter_path]
if feature == "pathways":
cmd += ["-m", paths.data_kegg_path("ensg_kegg.tsv")]
cmd += [ofm["data"]]
project["oncodrivefm"] = dict(
feature=feature,
slice=slice_name,
results=os.path.join(project["temp_path"], "oncodrivefm-{0}-{1}.tsv".format(feature, slice_name)))
cmd = " ".join(cmd)
log.debug(cmd)
ret_code = subprocess.call(cmd, shell=True)
if ret_code != 0:
raise Exception("OncodriveFM error while computing {0}:\n{1}".format(feature, cmd))
projects_out_port.send(project)
def drivers():
log = task.logger
config = GlobalConfig(task.conf)
paths = PathsConfig(config)
db_path = paths.results_path("drivers.db")
db = SigDb(db_path)
db.open()
log.info("Variants ...")
path = paths.combination_path("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 = paths.combination_path("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 = paths.combination_path("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 = paths.results_path("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()
def combination_oncodrivefm(projects_set):
log = task.logger
config = GlobalConfig(task.conf)
paths = PathsConfig(config)
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 = paths.combination_path("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 fimpact_run(partition):
log = task.logger
config = GlobalConfig(task.conf)
paths = PathsConfig(config)
results_port = task.ports("results")
project = partition["project"]
log.info("--- [{0} @ {1}] --------------------------------------------".format(project["id"], partition["index"]))
log.info("Reading MA scores ...")
ma_uniprot = {}
ma_scores = {}
with open(partition["ma_path"], "r") as f:
for var_id, uniprot, fi_score in tsv.lines(f, (int, str, float), null_value="-"):
ma_uniprot[var_id] = uniprot
ma_scores[var_id] = fi_score
log.info("Reading VEP results and calculating functional impact ...")
tfic = TransFIC(data_path=paths.data_transfic_path())
tfi_path = os.path.join(partition["base_path"], "{0:08d}.tfi".format(partition["index"]))
cf = open(tfi_path, "w")
with open(partition["vep_path"], "r") as f:
for fields in tsv.lines(f, (int, str, str, str, str, str, str, float, float), null_value="-"):
(var_id, gene, transcript, ct,
protein_pos, aa_change, protein,
sift_score, pph2_score) = fields
ct = (ct or "").split(",")
# Invert sift score
if sift_score is not None:
sift_score = 1.0 - sift_score
ma_score = None
uniprot = ma_uniprot.get(var_id)
sift_impact = pph2_impact = ma_impact = None # TransFIC.UNKNOWN_IMPACT_CLASS
coding_region = 1 if so.match(ct, so.CODING_REGION) else 0
sift_tfic, sift_class, pph2_tfic, pph2_class, ma_tfic, ma_class = (None, None, None, None, None, None)
ct_type = None
if so.match(ct, so.NON_SYNONYMOUS): # missense
ct_type = TransFIC.CT_NON_SYNONYMOUS
ma_score = ma_scores.get(var_id)
(sift_tfic, sift_class,
pph2_tfic, pph2_class,
ma_tfic, ma_class) = tfic.calculate("gosmf", gene, ct_type, sift_score, pph2_score, ma_score)
sift_impact = sift_class if sift_class in IMPACT_CLASSES else sift_impact
pph2_impact = pph2_class if pph2_class in IMPACT_CLASSES else pph2_impact
ma_impact = ma_class if ma_class in IMPACT_CLASSES else ma_impact
elif so.match(ct, so.STOP): # stop
sift_impact = pph2_impact = ma_impact = TransFIC.HIGH_IMPACT_CLASS
sift_score = pph2_score = 1.0
ma_score = 3.5
elif so.match(ct, so.FRAMESHIFT): # frameshift
sift_impact = pph2_impact = ma_impact = TransFIC.HIGH_IMPACT_CLASS
sift_score = pph2_score = 1.0
ma_score = 3.5
elif so.match(ct, so.SPLICE_JUNCTION): # splice junction
sift_impact = pph2_impact = ma_impact = TransFIC.HIGH_IMPACT_CLASS
sift_score = pph2_score = 1.0
ma_score = 3.5
elif so.match(ct, so.SPLICE_REGION): # splice region
sift_impact = pph2_impact = ma_impact = TransFIC.UNKNOWN_IMPACT_CLASS
sift_score = pph2_score = 1.0
ma_score = 3.5
elif so.match(ct, so.SYNONYMOUS): # synonymous
sift_impact = pph2_impact = ma_impact = TransFIC.NONE_IMPACT_CLASS
sift_score = pph2_score = 0.0
ma_score = -2
else:
sift_impact = pph2_impact = ma_impact = TransFIC.NONE_IMPACT_CLASS
aff_gene = (var_id, gene)
# try to follow the convention http://www.hgvs.org/mutnomen/recs-prot.html
prot_change = None
if ct_type == TransFIC.CT_FRAMESHIFT:
if protein_pos is None:
prot_change = "fs"
else:
prot_change = "fs {0}".format(protein_pos)
#log.debug("FRAMESHIFT: gene={}, protein={}, pos={}, change={}".format(gene, protein, protein_pos, aa_change))
elif ct_type == "splice":
prot_change = "r.spl?"
#log.debug("SPLICE: gene={}, protein={}, pos={}, change={}".format(gene, protein, protein_pos, aa_change))
elif protein_pos is not None and aa_change is not None:
rc = ReContext()
if rc.match(SIMPLE_AA_CHANGE_RE, aa_change):
prot_change = "{ref}{pos}{alt}".format(pos=protein_pos, ref=rc.group(1), alt=rc.group(2) or "=")
elif rc.match(COMPLEX_AA_CHANGE_RE, aa_change):
prot_change = "{0} {1}".format(aa_change, protein_pos)
else:
log.warn("Unmatched aa change: gene={}, protein={}, pos={}, change={}, ct=[{}]".format(
gene, protein, protein_pos, aa_change, ", ".join(ct)))
tr_impact = ma_impact or pph2_impact or sift_impact or TransFIC.UNKNOWN_IMPACT_CLASS
tsv.write_line(cf, var_id, transcript, gene, uniprot, prot_change, coding_region, tr_impact,
sift_score, sift_tfic, sift_class, sift_impact,
pph2_score, pph2_tfic, pph2_class, pph2_impact,
ma_score, ma_tfic, ma_class, ma_impact,
null_value="-")
cf.close()
# Send results to the next module
partition["tfi_path"] = tfi_path
results_port.send(partition)
def begin():
log = task.logger
config = GlobalConfig(task.conf)
paths = PathsConfig(config)
log.info("Creating combination folders ...")
paths.create_combination_folders()
log.info("Checking classifiers ...")
"""
if CONF_KEY in conf:
classifiers = conf[CONF_KEY].to_native()
else:
classifiers = []
"""
classifiers = config.combination.classifiers
results = []
if len(classifiers) == 0:
log.warn("No classifiers have been defined !!!")
log.warn("Specify them in the configuration with '{0}'".format(CONF_KEY))
updated_classifiers = []
for index, classifier in enumerate(classifiers):
classifier = classifier.to_native()
if not isinstance(classifier, dict):
raise Exception("Classifier at index {0} should be a dictionary".format(index))
if "id" not in classifier:
classifier["id"] = str(index)
classifier_id = classifier["id"]
if "name" not in classifier:
classifier["name"] = "Classifier {0}".format(index)
if "keys" not in classifier:
classifier["keys"] = []
elif not isinstance(classifier["keys"], list):
raise Exception("'keys' for classifier at index {0} should be a list".format(classifier_id))
keys_len = len(classifier["keys"])
if "default_key_values" not in classifier:
classifier["default_key_values"] = [""] * keys_len
elif not isinstance(classifier["default_key_values"], list):
raise Exception("'default_key_values' for classifier {0} should be a list".format(classifier_id))
elif len(classifier["default_key_values"]) != keys_len:
raise Exception("Number of values for 'default_key_values' should be the same of 'keys' in classifier '{0}'".format(classifier_id))
if "short_names" not in classifier:
classifier["short_names"] = classifier["keys"]
elif not isinstance(classifier["short_names"], list):
raise Exception("'short_names' for classifier {0} should be a list".format(classifier_id))
elif len(classifier["short_names"]) != keys_len:
raise Exception("Number of keys for 'short_names' should be the same of 'keys' in classifier '{0}'".format(classifier_id))
if "default_short_values" not in classifier:
classifier["default_short_values"] = classifier["default_key_values"]
elif not isinstance(classifier["default_short_values"], list):
raise Exception("'default_short_values' for classifier {0} should be a list".format(classifier_id))
elif len(classifier["default_short_values"]) != keys_len:
raise Exception("Number of values for 'default_short_values' should be the same of 'keys' in classifier '{0}'".format(classifier_id))
if "long_names" not in classifier:
classifier["long_names"] = classifier["short_names"]
elif not isinstance(classifier["long_names"], list):
raise Exception("'long_names' for classifier {0} should be a list".format(classifier_id))
elif len(classifier["long_names"]) != keys_len:
raise Exception("Number of keys for 'long_names' should be the same of 'keys' in classifier '{0}'".format(classifier["id"]))
if "default_long_values" not in classifier:
classifier["default_long_values"] = classifier["default_key_values"]
elif not isinstance(classifier["default_long_values"], list):
raise Exception("'default_long_values' for classifier {0} should be a list".format(classifier_id))
elif len(classifier["default_long_values"]) != keys_len:
raise Exception("Number of values for 'default_long_values' should be the same of 'keys' in classifier '{0}'".format(classifier_id))
updated_classifiers += [classifier]
results += [{}]
task.context["classifiers"] = updated_classifiers
task.context["results"] = results