def findVarDisGeneDrug(pmid, text):
"""
>>> startup({})
>>> list(findVarDisGeneDrug(0, "The R71G BRCA1 is a breast cancer founder mutation not treatable with Herceptin"))
"""
textLow = text.lower()
# very basic filter, remove documents without some basic keywords
if " variant " not in textLow and " mutation" not in textLow and " substitution" not in textLow and \
" mutant " not in textLow:
return
for section, sentStart, sentEnd, sentText in pubNlp.sectionSentences(text):
genes = list(geneFinder.findGeneNames(sentText))
if len(genes) == 0:
continue
#print "genes", genes, sentText
conds = list(pubNlp.findDiseases(sentText))
drugs = list(pubNlp.findDrugs(sentText))
# remove diseases and drugs that are also genes
drugs = pubNlp.rangeRemoveOverlaps(drugs, genes)
conds = pubNlp.rangeRemoveOverlaps(conds, genes)
# check if we still have a disease and drug left
if len(conds) == 0 or len(drugs) == 0:
continue
print "drugs", drugs
print "diseases", conds
geneSnips = pubNlp.rangeTexts(sentText, genes)
condSnips = pubNlp.rangeTexts(sentText, conds)
drugSnips = pubNlp.rangeTexts(sentText, drugs)
genePosSet = pubNlp.rangeToPosSet(genes)
variants = varFinder.findVariantDescriptions(sentText,
exclPos=genePosSet)
# the last field of the genes rows is the entrez ID
entrezIds = [r[-1] for r in genes]
# we need a protein variant, not DNA
if "prot" not in variants:
continue
for variant, mentions in variants["prot"]:
print "grounding variant", variant, mentions
groundedMuts, ungroundVar, beds = \
varFinder.groundVariant(pmid, sentText, variant, mentions, [], entrezIds)
for mutInfo in groundedMuts:
coords = [(m.start, m.end) for m in mentions]
varSnips = pubNlp.rangeTexts(sentText, coords)
row = [
section, "|".join(geneSnips), "|".join(condSnips),
"|".join(drugSnips), "|".join(varSnips), sentText
]
yield row
def findVarDisGeneDrug(pmid, text):
"""
>>> startup({})
>>> list(findVarDisGeneDrug(0, "The R71G BRCA1 is a breast cancer founder mutation not treatable with Herceptin"))
"""
textLow = text.lower()
# very basic filter, remove documents without some basic keywords
if " variant " not in textLow and " mutation" not in textLow and " substitution" not in textLow and \
" mutant " not in textLow:
return
for section, sentStart, sentEnd, sentText in pubNlp.sectionSentences(text):
genes = list(geneFinder.findGeneNames(sentText))
if len(genes)==0:
continue
#print "genes", genes, sentText
conds = list(pubNlp.findDiseases(sentText))
drugs = list(pubNlp.findDrugs(sentText))
# remove diseases and drugs that are also genes
drugs = pubNlp.rangeRemoveOverlaps(drugs, genes)
conds = pubNlp.rangeRemoveOverlaps(conds, genes)
# check if we still have a disease and drug left
if len(conds)==0 or len(drugs)==0:
continue
print "drugs", drugs
print "diseases", conds
geneSnips = pubNlp.rangeTexts(sentText, genes)
condSnips = pubNlp.rangeTexts(sentText, conds)
drugSnips = pubNlp.rangeTexts(sentText, drugs)
genePosSet = pubNlp.rangeToPosSet(genes)
variants = varFinder.findVariantDescriptions(sentText, exclPos=genePosSet)
# the last field of the genes rows is the entrez ID
entrezIds = [r[-1] for r in genes]
# we need a protein variant, not DNA
if "prot" not in variants:
continue
for variant, mentions in variants["prot"]:
print "grounding variant", variant, mentions
groundedMuts, ungroundVar, beds = \
varFinder.groundVariant(pmid, sentText, variant, mentions, [], entrezIds)
for mutInfo in groundedMuts:
coords = [(m.start, m.end) for m in mentions]
varSnips = pubNlp.rangeTexts(sentText, coords)
row = [section, "|".join(geneSnips), "|".join(condSnips), "|".join(drugSnips), "|".join(varSnips), sentText]
yield row
def annotateFile(article, file):
" go over words of text and check if they are in dict "
text = file.content
if len(text)>100000:
return
if reqStrings!=None:
found = False
#sentLower = sent.lower()
textLower = text.lower()
for rs in reqStrings:
if text.find(rs)!=-1:
#if sentLower.find(rs)!=-1:
found = True
break
if not found:
return
for section, sentStart, sentEnd, sent in pubNlp.sectionSentences(text, file.fileType, mustHaveVerb=False):
#if len(sent)<20:
#logging.debug("Sentence too short: %d characters" % len(text))
#continue
#if len(sent)>2000:
#logging.debug("Sentence too long: %d characters" % len(text))
#continue
found = True
posList = []
allMatches = []
for lexName, lex in lexes.iteritems():
matches = []
lexMatches = fastFind(sent, lex, toLower=toLower)
if len(lexMatches)==0 or len(lexMatches)>10:
found = False
break
for start, end, word in lexMatches:
matches.append(word.replace("="," ").replace(",", " "))
posList.append("%d-%d" % (start, end))
allMatches.append("%s=%s" % (lexName, ",".join(matches)))
if found:
yield sentStart, sentEnd, "|".join(allMatches), "|".join(posList), sent
def annotateFile(article, file):
" go over words of text and check if they are in dict "
text = file.content
if len(text) > 100000:
return
if reqStrings != None:
found = False
# sentLower = sent.lower()
textLower = text.lower()
for rs in reqStrings:
if text.find(rs) != -1:
# if sentLower.find(rs)!=-1:
found = True
break
if not found:
return
for section, sentStart, sentEnd, sent in pubNlp.sectionSentences(text, file.fileType, mustHaveVerb=False):
# if len(sent)<20:
# logging.debug("Sentence too short: %d characters" % len(text))
# continue
# if len(sent)>2000:
# logging.debug("Sentence too long: %d characters" % len(text))
# continue
found = True
posList = []
allMatches = []
for lexName, lex in lexes.iteritems():
matches = []
lexMatches = fastFind(sent, lex, toLower=toLower)
if len(lexMatches) == 0 or len(lexMatches) > 10:
found = False
break
for start, end, word in lexMatches:
matches.append(word.replace("=", " ").replace(",", " "))
posList.append("%d-%d" % (start, end))
allMatches.append("%s=%s" % (lexName, ",".join(matches)))
if found:
yield sentStart, sentEnd, "|".join(allMatches), "|".join(posList), sent
def annotateFile(self, article, file):
text = file.content
pmid = article.pmid
if file.fileType == "supp":
return
for row in pubNlp.sectionSentences(text, file.fileType):
section, sentStart, sentEnd, text = row
tokens = text.split()
if len(tokens) < 6:
logging.debug("Sentence too short: %d tokens" % len(tokens))
continue
if len(tokens) > 40:
logging.debug("Sentence too long: %d tokens" % len(tokens))
continue
if len(text) < 20:
logging.debug("Sentence too short: %d characters" % len(text))
continue
if len(text) > 1000:
logging.debug("Sentence too long: %d characters" % len(text))
continue
if text.count('"') > 20 or text.count(",") > 20:
logging.debug("Too many strange characters")
continue
genes = list(geneFinder.findGeneNames(text))
if len(genes) < 2:
continue
if len(genes) > 20:
logging.debug("Too many genes, %d" % len(genes))
continue
geneDescs = ["%d-%d/%s/%s/%s" % (start,end,text[start:end],name,gid) \
for start,end,name,gid in genes]
geneDesc = "|".join(geneDescs)
row = [pmid, section, start, end, text, geneDesc]
yield row
self.rowCount += 1
if self.rowCount % 200 == 0:
yield [] # tell caller to start a new output file
def annotateFile(self, article, file):
text = file.content
pmid = article.pmid
if file.fileType=="supp":
return
for row in pubNlp.sectionSentences(text, file.fileType):
section, sentStart, sentEnd, text = row
tokens = text.split()
if len(tokens)<6:
logging.debug("Sentence too short: %d tokens" % len(tokens))
continue
if len(tokens)>40:
logging.debug("Sentence too long: %d tokens" % len(tokens))
continue
if len(text)<20:
logging.debug("Sentence too short: %d characters" % len(text))
continue
if len(text)>1000:
logging.debug("Sentence too long: %d characters" % len(text))
continue
if text.count('"') > 20 or text.count(",")>20:
logging.debug("Too many strange characters")
continue
genes = list(geneFinder.findGeneNames(text))
if len(genes) < 2:
continue
if len(genes) > 20:
logging.debug("Too many genes, %d" % len(genes))
continue
geneDescs = ["%d-%d/%s/%s/%s" % (start,end,text[start:end],name,gid) \
for start,end,name,gid in genes]
geneDesc = "|".join(geneDescs)
row = [pmid, section, start, end, text, geneDesc]
yield row
self.rowCount += 1
if self.rowCount % 200 == 0:
yield [] # tell caller to start a new output file
def findDisGeneVariant(text):
"""
>>> geneFinder.initData(exclMarkerTypes=["dnaSeq", "band"])
>>> varFinder.loadDb(loadSequences=False)
>>> list(findDisGeneVariant("Diabetes is caused by a PITX2 mutation, V234T and influenced by Herceptin."))
[(0, 74, 'probablyAbstract', '64-73:Herceptin=Trastuzumab', '0-8:Diabetes=Diabetes Mellitus', '24-29:PITX2=symbol', 'V233T', 'Diabetes is caused by a PITX2 mutation, V234T and influenced by Herceptin.')]
>>> #list(findDisGeneVariant("We undertook a quantitative review of the literature to estimate the effectiveness of desferrioxamine and deferiprone in decreasing hepatic iron concentrations (HIC) in thalassemia major."))
>>> list(findDisGeneVariant("his mutation, we cotransfected C3H10T cells with expression vectors encoding SMO-WT or SMO-D473H "))
"""
docGenes = list(geneFinder.findGeneNames(text))
docEntrezIds = set([r[-1] for r in docGenes])
for section, start, end, sentence in pubNlp.sectionSentences(text):
conds = list(pubNlp.findDiseases(sentence))
drugs = list(pubNlp.findDrugs(sentence))
genes = list(geneFinder.findGeneNames(sentence))
#print conds, drugs, genes, section, sentence
# remove drugs and conds that are also genes
drugs = rangeRemoveOverlaps(drugs, genes)
conds = rangeRemoveOverlaps(conds, genes)
#geneSnips = rangeTexts(sentence, genes, useSym=True)
#condSnips = rangeTexts(sentence, conds)
#drugSnips = rangeTexts(sentence, drugs)
#
mutDescs = []
mutDict = varFinder.findVariantDescriptions(sentence)
if "prot" in mutDict:
for varDesc, mentions in mutDict["prot"]:
if varDesc.mutType!="sub":
continue
logging.debug("grounding variant: %s %s"% (varDesc, mentions))
groundedMuts, ungroundVar, beds = \
varFinder.groundVariant(None, sentence, varDesc, mentions, [], docEntrezIds)
for mutInfo in groundedMuts:
shortDesc = varDesc.origSeq+str(varDesc.start+1)+varDesc.mutSeq # 0-based!!
mutDescs.append(shortDesc+"=%s:%s"%(mutInfo.geneSymbol,mutInfo.hgvsProt))
#mutMatches = list(mutRe.finditer(sentence))
#mutDescs = [(m.group(1),m.group(2), m.group(3)) for m in mutMatches]
#mutDescSet = set(mutDescs)
#blackListMuts = mutDescSet.intersection(blackListStr)
#if len(mutMatches)==0:
#logging.debug("No mutation found, skipping")
#continue
#if len(blackListMuts)!=0:
#logging.debug("At least one blacklisted mutation found, skipping")
#continue
#if len(drugs)==0:
#logging.debug("No drugs found, skipping")
#continue
#if len(genes)==0:
#logging.debug("No genes found, skipping")
#continue
mutDesc = "|".join(mutDescs)
drugDesc = rangeDescs(sentence, drugs)
condDesc = rangeDescs(sentence, conds)
geneDesc = rangeDescs(sentence, genes, useSym=True)
ret = (start, end, section, drugDesc, condDesc, geneDesc, mutDesc, sentence)
yield ret
def findDisGeneVariant(text):
"""
>>> geneFinder.initData(exclMarkerTypes=["dnaSeq", "band"])
>>> varFinder.loadDb(loadSequences=False)
>>> list(findDisGeneVariant("Diabetes is caused by a PITX2 mutation, V234T and influenced by Herceptin."))
[(0, 74, 'probablyAbstract', '64-73:Herceptin=Trastuzumab', '0-8:Diabetes=Diabetes Mellitus', '24-29:PITX2=symbol', 'V233T', 'Diabetes is caused by a PITX2 mutation, V234T and influenced by Herceptin.')]
>>> #list(findDisGeneVariant("We undertook a quantitative review of the literature to estimate the effectiveness of desferrioxamine and deferiprone in decreasing hepatic iron concentrations (HIC) in thalassemia major."))
>>> list(findDisGeneVariant("his mutation, we cotransfected C3H10T cells with expression vectors encoding SMO-WT or SMO-D473H "))
"""
docGenes = list(geneFinder.findGeneNames(text))
docEntrezIds = set([r[-1] for r in docGenes])
for section, start, end, sentence in pubNlp.sectionSentences(text):
conds = list(pubNlp.findDiseases(sentence))
drugs = list(pubNlp.findDrugs(sentence))
genes = list(geneFinder.findGeneNames(sentence))
#print conds, drugs, genes, section, sentence
# remove drugs and conds that are also genes
drugs = rangeRemoveOverlaps(drugs, genes)
conds = rangeRemoveOverlaps(conds, genes)
#geneSnips = rangeTexts(sentence, genes, useSym=True)
#condSnips = rangeTexts(sentence, conds)
#drugSnips = rangeTexts(sentence, drugs)
#
mutDescs = []
mutDict = varFinder.findVariantDescriptions(sentence)
if "prot" in mutDict:
for varDesc, mentions in mutDict["prot"]:
if varDesc.mutType != "sub":
continue
logging.debug("grounding variant: %s %s" % (varDesc, mentions))
groundedMuts, ungroundVar, beds = \
varFinder.groundVariant(None, sentence, varDesc, mentions, [], docEntrezIds)
for mutInfo in groundedMuts:
shortDesc = varDesc.origSeq + str(
varDesc.start + 1) + varDesc.mutSeq # 0-based!!
mutDescs.append(shortDesc + "=%s:%s" %
(mutInfo.geneSymbol, mutInfo.hgvsProt))
#mutMatches = list(mutRe.finditer(sentence))
#mutDescs = [(m.group(1),m.group(2), m.group(3)) for m in mutMatches]
#mutDescSet = set(mutDescs)
#blackListMuts = mutDescSet.intersection(blackListStr)
#if len(mutMatches)==0:
#logging.debug("No mutation found, skipping")
#continue
#if len(blackListMuts)!=0:
#logging.debug("At least one blacklisted mutation found, skipping")
#continue
#if len(drugs)==0:
#logging.debug("No drugs found, skipping")
#continue
#if len(genes)==0:
#logging.debug("No genes found, skipping")
#continue
mutDesc = "|".join(mutDescs)
drugDesc = rangeDescs(sentence, drugs)
condDesc = rangeDescs(sentence, conds)
geneDesc = rangeDescs(sentence, genes, useSym=True)
ret = (start, end, section, drugDesc, condDesc, geneDesc, mutDesc,
sentence)
yield ret
