def filterNetMHC(allele,
length,
baseScoreTable,
baseChainCollection,
baseFasta,
additionalScoreTable,
additionalChainCollection,
additionalFasta,
k,
reverse=False):
baseScoreDist = collections.Counter()
if baseScoreTable and baseChainCollection and baseFasta:
exclusion = None
if additionalChainCollection:
additionalGen = getPeptideGen(additionalChainCollection,
additionalFasta, length)
exclusion = set(additionalGen)
baseGen = getPeptideGen(baseChainCollection, baseFasta, length)
assert (allele in baseScoreTable.getAlleles())
l = list(baseScoreTable.scoreIter(allele))
baseScoreDist = scoreDistribution(l, baseGen, exclusion)
additionalScoreDist = collections.Counter()
if additionalScoreTable and additionalChainCollection and additionalFasta:
additionalGen = getPeptideGen(additionalChainCollection,
additionalFasta, length)
additionalScoreDist = scoreDistribution(
additionalScoreTable.scoreIter(allele), additionalGen, None)
combinedScoreDist = additionalScoreDist + baseScoreDist
threshold = computeScoreThreshold(combinedScoreDist, k, reverse)
peptides = []
pepToHeader = collections.defaultdict(set)
if baseScoreTable and baseChainCollection and baseFasta:
gen = peptideGenerator(baseChainCollection, baseFasta, length)
pep = filterOnThreshold(baseScoreTable.scoreIter(allele), gen,
threshold, reverse)
for holder in pep:
pepToHeader[holder.getPeptideSequence()].update(
holder.getHeaders())
if additionalScoreTable and additionalChainCollection and additionalFasta:
gen = peptideGenerator(additionalChainCollection, additionalFasta,
length)
pep = filterOnThreshold(additionalScoreTable.scoreIter(allele), gen,
threshold, reverse)
for holder in pep:
pepToHeader[holder.getPeptideSequence()].update(
holder.getHeaders())
return pepToHeader
def getPeptideGen(chainCollection, fastaPath, pepLen):
return map(lambda x: x.getPeptideSequence(),
peptideGenerator(chainCollection, fastaPath, pepLen))
assert(bindingScoreTable.peptideLength == chainCollection.peptideLength)
if args.allele in bindingScoreTable.getAlleles():
print('Allele already in binding score table. Exiting')
sys.exit(1)
def getPeptideGen(chainCollection, fastaPath, pepLen):
return map(lambda x: x.getPeptideSequence(), peptideGenerator(chainCollection, fastaPath, pepLen))
predictor=mhcflurry.Class1AffinityPredictor.load()
KNOWN_AMINOS=set('ACDEFGHIKLMNPQRSTVWY')
def unknownSub(peptides):
pep = []
for x in peptides:
sub = ''.join([y if y in KNOWN_AMINOS else 'X' for y in x])
pep.append(sub)
return pep
peptides = [x.getPeptideSequence() for x in peptideGenerator(chainCollection, args.fasta, args.length)]
peptidesWithUnknown = unknownSub(peptides)
scores = list(predictor.predict(allele=args.allele, peptides=peptidesWithUnknown))
result = bindingScoreTable.addAllele(args.allele, iter(scores))
parser.add_argument('chains')
parser.add_argument('scoreTable')
parser.add_argument('allele')
parser.add_argument('k', type=int)
args = parser.parse_args()
assert (os.path.isfile(args.chains))
assert (os.path.isfile(args.fasta))
assert (os.path.isfile(args.scoreTable))
assert (args.k > 0)
chainCollection = None
with open(args.chains, 'rb') as f:
chainCollection = pickle.load(f)
f = open(args.scoreTable, 'rb')
st = ScoreTable.readExisting(f)
print('alleles: ')
print(st.getAlleles())
assert (args.allele in st.getAlleles())
length = st.peptideLength
n = st.numPeptides
assert (args.k <= n)
print('num peptides: ' + str(n))
pepGen = peptideGenerator(chainCollection, args.fasta, length)
selection = sorted(random.sample(range(0, n), args.k))
scoreIter = st.scoreIter(args.allele)
pepAndScores = extractPeptidesAndScores(scoreIter, pepGen, selection)
for pep, score in pepAndScores:
print(str(pep) + '\t' + str(score))