def annotateFile(self, article, file):
" go over words of text and check if they are in dict "
text = file.content
count = 0
rows = []
for start, end, markerType, markerId in geneFinder.findIdentifiers(
text):
# never resolve some types
if markerType in ["band", "genbank", "refseq", "ensembl", "snp"]:
if markerType == "snp":
markerId = "rs" + markerId
row = [start, end, markerType, markerId, "", ""]
rows.append(row)
else:
# try to resolve most other types to genes
genes = geneFinder.markerToGenes(markerType, markerId)
if genes != None and len(genes) < MAXGENEPERID:
geneSyms = set(genes)
for geneSym in geneSyms:
geneId, geneSym = genes.items()[0]
row = [
start, end, "gene", geneSym, markerType, markerId
]
rows.append(row)
else:
row = [start, end, markerType, markerId, "", ""]
rows.append(row)
if len(rows) > MAXCOUNT:
logging.info("%d annotations, too mant, skipping file %s" %
(MAXCOUNT, file.externalId))
return None
return rows
def map(self, article, file, text, resultDict):
matches = list(geneFinder.findIdentifiers(text))
for start, end, type, word in matches:
resultDict.setdefault(word, set()).add(file.fileId)