def check_ORNA_in_ago(oFN, oFF, agoFN, clippingAmount = 1):
NX = Nexus(oFN, oFF)
NX.load(['sequence', 'geneNames'])
#make truncated sequences
id_sequence = NX.createMap('id', 'sequence')
if clippingAmount > 0:
id_sequence = dict( (i, j[clippingAmount:-clippingAmount]) for i,j in id_sequence.items())
#get fastq sequences
agoF = open(agoFN, 'r')
agoSeqs = []
while True:
fPacket = nextFilePacket(agoF, 4)
if not fPacket: break
agoSeqs.append(fPacket[1])
agoF.close()
#count for each oRNA
id_count = {}
for id, seq in id_sequence.items():
for agoSeq in agoSeqs:
if seq in agoSeq:
id_count[id] = id_count.get(id, 0) + 1
#out
totalCount = 0
for id, count in id_count.items():
NX.id = id
print '%s\t%s\t%s' % (id, count, NX.geneNames)
totalCount += count
print totalCount
def check_ORNA_in_ago(oFN, oFF, agoFN, clippingAmount=1):
NX = Nexus(oFN, oFF)
NX.load(['sequence', 'geneNames'])
#make truncated sequences
id_sequence = NX.createMap('id', 'sequence')
if clippingAmount > 0:
id_sequence = dict((i, j[clippingAmount:-clippingAmount])
for i, j in id_sequence.items())
#get fastq sequences
agoF = open(agoFN, 'r')
agoSeqs = []
while True:
fPacket = nextFilePacket(agoF, 4)
if not fPacket: break
agoSeqs.append(fPacket[1])
agoF.close()
#count for each oRNA
id_count = {}
for id, seq in id_sequence.items():
for agoSeq in agoSeqs:
if seq in agoSeq:
id_count[id] = id_count.get(id, 0) + 1
#out
totalCount = 0
for id, count in id_count.items():
NX.id = id
print '%s\t%s\t%s' % (id, count, NX.geneNames)
totalCount += count
print totalCount
def updateSimilarSiblings(oFN, oFF, frameLength):
dataNX = Nexus(oFN, oFF)
dataNX.load(['sequence', 'siblingSet'])
oID_sequence = dataNX.createMap('id', 'sequence')
consolidatedSets = getSimilarORNASets(oID_sequence, frameLength)
for cSet in consolidatedSets:
for oID in cSet:
dataNX.id = oID
dataNX.siblingSet = list(cSet)
dataNX.save()
def updateSimilarSiblings(oFN, oFF, frameLength):
dataNX = Nexus(oFN, oFF)
dataNX.load(['sequence', 'siblingSet'])
oID_sequence = dataNX.createMap('id', 'sequence')
consolidatedSets = getSimilarORNASets(oID_sequence, frameLength)
for cSet in consolidatedSets:
for oID in cSet:
dataNX.id = oID
dataNX.siblingSet = list(cSet)
dataNX.save()
def updateTargetIDs(oFN, oFF, aFN, aFF):
NX = Nexus(oFN, oFF)
NX.load(['filteredTargets'])
aNX = Nexus(aFN, aFF)
aNX.load(['sID'])
while aNX.nextID():
NX.id = aNX.sID
NX.filteredTargets.append(aNX.id)
NX.save()
def linkTargetIDs(oFN, oFF, aFN, aFF):
oNX = Nexus(oFN, oFF)
oNX.load(['filteredTargets'])
#just give it some blanks
if os.path.getsize(aFN) == 0:
while oNX.nextID():
oNX.filteredTargets = []
oNX.save()
return
aNX = Nexus(aFN, aFF)
aNX.load(['sID'])
sID_aIDs = aNX.createMap('sID', 'id', False)
for sID, aIDs in sID_aIDs.iteritems():
oNX.id = sID
oNX.filteredTargets = aIDs
oNX.save()
def linkTargetIDs(oFN, oFF, aFN, aFF):
oNX = Nexus(oFN, oFF)
oNX.load(['filteredTargets'])
#just give it some blanks
if os.path.getsize(aFN) == 0:
while oNX.nextID():
oNX.filteredTargets = []
oNX.save()
return
aNX = Nexus(aFN, aFF)
aNX.load(['sID'])
sID_aIDs = aNX.createMap('sID', 'id', False)
for sID, aIDs in sID_aIDs.iteritems():
oNX.id = sID
oNX.filteredTargets = aIDs
oNX.save()
def updateNumUFBS(oFN, oFF, aFN, aFF):
oNX = Nexus(oFN, oFF)
oNX.load(['filteredTargets', 'numUFBS'])
#just give it some blanks
if os.path.getsize(aFN) == 0:
while oNX.nextID():
oNX.numUFBS = 0
oNX.save()
return
aNX = Nexus(aFN, aFF)
aNX.load(['sigMask'])
while oNX.nextID():
sigMaskSet = set()
for aID in oNX.filteredTargets:
aNX.id = aID
sigMaskSet.add(aNX.sigMask)
oNX.numUFBS = len(sigMaskSet)
oNX.save()
def updateNumUFBS(oFN, oFF, aFN, aFF):
oNX = Nexus(oFN, oFF)
oNX.load(['filteredTargets', 'numUFBS'])
#just give it some blanks
if os.path.getsize(aFN) == 0:
while oNX.nextID():
oNX.numUFBS = 0
oNX.save()
return
aNX = Nexus(aFN, aFF)
aNX.load(['sigMask'])
while oNX.nextID():
sigMaskSet = set()
for aID in oNX.filteredTargets:
aNX.id = aID
sigMaskSet.add(aNX.sigMask)
oNX.numUFBS = len(sigMaskSet)
oNX.save()
