def get_transcripts(fanns_db, mut_cds, mut_aa, mut_pos, mut_strand, acc, logger):
fields = [mut_cds, mut_aa, mut_pos, mut_strand, acc]
cds_ctx = ReContext(mut_cds)
aa_ctx = ReContext(mut_aa)
if cds_ctx.match(MUT_CDS_RE):
ref, alt = [cds_ctx.group(i) for i in xrange(1, 3)]
aa_ref_len = len(ref)
if aa_ref_len != len(alt):
logger.warn("Found substitution with different alleles: {}".format(fields))
if mut_strand == "-":
ref = complementary_sequence(ref)
alt = complementary_sequence(alt)
pos_ctx = ReContext(mut_pos)
if not pos_ctx.match(MUT_POS_RE):
logger.warn("Unexpected mutation position: {}".format(fields))
return
chrom, start = [pos_ctx.group(i) for i in xrange(1, 3)]
if chrom == "25":
return
start = int(start)
for i in xrange(aa_ref_len):
#logger.info("{}{}:{}:{}/{}:{} ({}, {})".format(chrom, mut_strand, start+i, ref[i], alt[i], acc, mut_cds, mut_aa))
for row in fanns_db.query_scores(chr=chrom, start=start + i, ref=ref[i], alt=alt[i],
strand=mut_strand, transcript=acc, maps=["symbol"]):
#logger.info(" -> {}".format(row))
yield row
elif aa_ctx.match(MUT_AA_RE):
aa_ref, aa_pos, aa_alt = [aa_ctx.group(i) for i in xrange(1, 4)]
aa_ref_len = len(aa_ref)
if aa_ref_len != len(aa_alt):
logger.warn("Found substitution with different alleles: {}".format(fields))
aa_pos = int(aa_pos)
for i in xrange(aa_ref_len):
for row in fanns_db.query_scores(protein=acc, aa_pos=aa_pos + i, aa_ref=aa_ref[i], aa_alt=aa_alt[i],
maps=["symbol", "prot_transcript"]):
yield row
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 fimpact_run(partition):
log = task.logger
conf = task.conf
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=os.path.join(conf["data_path"], "TransFIC"))
tfi_path = os.path.join(partition["base_path"], "{0:08d}.tfi".format(partition["index"]))
cf = open(tfi_path, "w")
aff_gene_attrs = {}
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
if ct is not None:
ct = ct.split(",")
else:
ct = []
# Invert sift score
if sift_score is not None:
sift_score = 1.0 - sift_score
ma_score = None
uniprot = ma_uniprot[var_id] if var_id in ma_uniprot else None
sift_impact = pph2_impact = ma_impact = None # TransFIC.UNKNOWN_IMPACT_CLASS
coding_region = so.match(ct, so.CODING_REGION)
calculate_transfic = True
ct_type = None
if so.match(ct, so.NON_SYNONYMOUS): # missense
ct_type = TransFIC.CT_NON_SYNONYMOUS
ma_score = ma_scores[var_id] if var_id in ma_scores else None
elif so.match(ct, so.STOP): # stop
ct_type = TransFIC.CT_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
ct_type = TransFIC.CT_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): # splice
ct_type = "splice"
sift_impact = pph2_impact = ma_impact = TransFIC.HIGH_IMPACT_CLASS if so.match(ct, so.SPLICE_JUNCTION) else TransFIC.UNKNOWN_IMPACT_CLASS
calculate_transfic = False
elif so.match(ct, so.SYNONYMOUS): # synonymous
ct_type = TransFIC.CT_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
calculate_transfic = False
if calculate_transfic:
(sift_tfic, sift_class,
pph2_tfic, pph2_class,
ma_tfic, ma_class) = tfic.calculate("gosmf", gene, ct_type, sift_score, pph2_score, ma_score)
# if the impact was not preassigned get it from the transFIC calculated class
sift_impact = sift_class if sift_impact is None and sift_class in IMPACT_CLASSES else sift_impact
pph2_impact = pph2_class if pph2_impact is None and pph2_class in IMPACT_CLASSES else pph2_impact
ma_impact = ma_class if ma_impact is None and ma_class in IMPACT_CLASSES else ma_impact
else:
sift_tfic, sift_class, pph2_tfic, pph2_class, ma_tfic, ma_class = (None, None, None, None, None, None)
aff_gene = (var_id, gene)
# update aggregated impact for all the predictors
update_attr(aff_gene_attrs, aff_gene, "sift_impact", sift_impact, update=TransFIC.higher_impact)
update_attr(aff_gene_attrs, aff_gene, "pph2_impact", pph2_impact, update=TransFIC.higher_impact)
update_attr(aff_gene_attrs, aff_gene, "ma_impact", ma_impact, update=TransFIC.higher_impact)
# update whether the affected gene is a coding region or not
update_attr(aff_gene_attrs, aff_gene, "coding_region", coding_region,
update=lambda prev_value, value: prev_value or value)
# aggregate protein changes per affected_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)))
if prot_change is not None:
update_attr(aff_gene_attrs, aff_gene, "prot_changes", prot_change,
new=lambda value: set([value]),
update=lambda prev_value, value: prev_value | set([value]))
impact = ma_impact or pph2_impact or sift_impact or TransFIC.UNKNOWN_IMPACT_CLASS
tsv.write_line(cf, var_id, transcript, uniprot,
sift_score, sift_tfic, sift_class,
pph2_score, pph2_tfic, pph2_class,
ma_score, ma_tfic, ma_class,
impact, null_value="-")
cf.close()
log.info("Saving variant impacts ...")
gfi_path = os.path.join(partition["base_path"], "{0:08d}.gfi".format(partition["index"]))
vf = open(gfi_path, "w")
for aff_gene, attrs in aff_gene_attrs.items():
var_id, gene = aff_gene
# get the impact by trust priority: ma, pph2, sift
impact = attrs.get("ma_impact") or attrs.get("pph2_impact") or attrs.get("sift_impact") or TransFIC.UNKNOWN_IMPACT_CLASS
coding_region = attrs.get("coding_region", False)
coding_region = 1 if coding_region else 0
prot_changes = attrs.get("prot_changes")
prot_changes = ",".join(prot_changes) if prot_changes is not None else None
tsv.write_line(vf, var_id, gene, impact, coding_region, prot_changes, null_value="-")
vf.close()
# Send results to the next module
partition["tfi_path"] = tfi_path
partition["gfi_path"] = gfi_path
results_port.send(partition)
