def buildCtrlsFromMatchVers(self, restrictedList, numOfCtrls=15):
"""
WARNING: This should only be called after the entire list of real seeds have been initialized!
Computes usr supplied number of permutations (dflt=15) of each 'true' matchVersion and screens them
for real seqs in restrictedList to prevent using KNOWN seed matches for controls. Ctrl seed matches
are stored in a list located at second index of the list located in dict entry
self.matchVersions[seedType].
"""
# check to see whether this has already been done.
# If so, complain and die.
# Else, append an empty list as index_1 after REAL matchSeq for each version
for seedType in self.matchVersions:
assert type(self.matchVersions[seedType]) == type([]), \
'ERROR: %s.matchVersions[%s] is not type: list.' % (self.name,seedType)
assert len(self.matchVersions[seedType]) == 1, \
'ERROR: len(%s.matchVersions[%s]) is not 1; ctrls seqs may have already been built.' % (self.name,seedType)
self.matchVersions[seedType].append([])
# permute and screen each seedVersion
for seedType in self.matchVersions:
# If NO third index like here: [0,7,'A1']
if len(_seedModels[seedType]) == 2:
# Select numOfCtrls random permutations of matchVersions[seedType] that are not in the
# restrictedList.
matchPermList = [''.join(x) for x in xpermutations.xpermutations(list(self.matchVersions[seedType][0]))]
while len(self.matchVersions[seedType][1]) < numOfCtrls:
permSeq = JamesDefs.randFromList_noReplace(matchPermList)
if permSeq not in restrictedList:
# Append permuted Seq if not in restrictedList
self.matchVersions[seedType][1].append(permSeq)
# If extra data: use 'instructions index' to only permute the nucs not explicitly
# defined in the seedModel
elif len(_seedModels[seedType]) == 3:
nuc,pos = list(_seedModels[seedType][2])
# Leave 1-registered bc we will use negIndex bc dealing with rvCmp of miRNA
# so pos == 1 actually means pos == LAST in matchSeq
pos = int(pos)
# explode seq to remove defined nuc in place
seq2Perm = list(self.matchVersions[seedType][0])
del seq2Perm[-pos]
# Generate permutations from remaining nucs,
matchPermList = [x for x in xpermutations.xpermutations(seq2Perm)]
while len(self.matchVersions[seedType][1]) < numOfCtrls:
permSeq = JamesDefs.randFromList_noReplace(matchPermList)
# Replace nuc and check restricted list.
if pos > 1: permSeq.insert(-pos+1,nuc)
else: permSeq.append(nuc)
permSeq = ''.join(permSeq)
if permSeq not in restrictedList:
# Append permuted Seq if not in restrictedList
self.matchVersions[seedType][1].append(permSeq)
def shuffSeedsInMatMiRs(inPath,outPathBase,seed=[1,8],num=1):
"""Takes mature miR fasta file from inPath, permutes nucs in <seed>
<num> times and writes records to outPaths derived from outPathBase
with unique ctrl numbers.
<seed> coords are as pyhton slice coords (0-based and
non-inclusive on the right side)"""
inFile = open(inPath,'rU')
outFiles = []
for i in range(num):
outFiles.append(open('%s.ctrl%s.fas' % (outPathBase,i),'w'))
for line in inFile:
if line.startswith('>'):
for oF in outFiles:
oF.write(line)
else:
line = line.strip('\n')
seedSeq = line[seed[0]:seed[1]]
fvPrm = line[:seed[0]]
trPrm = line[seed[1]:]
shfSeeds = [x for x in xpermutations.xpermutations(list(seedSeq))]
sampShfs = sample(shfSeeds,num)
for i in range(num):
nLine = '%s%s%s\n' % (fvPrm,''.join(sampShfs[i]),trPrm)
outFiles[i].write(nLine)
def buildCtrlsFromProSeed(self, restrictedList, numOfCtrls=15):
"""
WARNING: This should only be called after the entire list of real seeds have been initialized!
Computes 15 permutations of the 'true' proSeed matching seqeunce (m2_to_m8) and derives
matchVersions as in the true case. The permuted sequence is checked agaisnt the restrictedList
to prevent using KNOWN seed matches for controls. Ctrl seed matches are stored in a list located
at second index of the list located in dict entry self.matchVersions[seedType]. Each seedVersion
of a ctrl set will share the same index number.
"""
##assert True==False, \
##"""WARNING!!! miRNA.buildCtrlsFromMatchVers() should be used instead!!
##If you REALLY want to use this method, modify or remove this assert statement.
##But seriously... use miRNA.buildCtrlsFromMatchVers()."""
# check to see whether this has already been done.
# If so, complain and die.
# Else, append an empty list as index_1 after REAL matchSeq for each version
for seedType in self.matchVersions:
assert type(self.matchVersions[seedType]) == type([]), \
'ERROR: %s.matchVersions[%s] is not type: list.' % (self.name,seedType)
assert len(self.matchVersions[seedType]) == 1, \
'ERROR: len(%s.matchVersions[%s]) is not 1; ctrls seqs may have already been built.' % (self.name,seedType)
self.matchVersions[seedType].append([])
proSeed = self.sourceSeq[1:8]
matchPerms = [''.join(x) for x in xpermutations.xpermutations(list(self.matchVersions['m2_to_m8'][0]))]
# Select 15 random permutations of matchVersions['m2_to_m8'] that are not in the
# restrictedList.
chosenPerms = []
while len(chosenPerms) < numOfCtrls:
permSeq = JamesDefs.randFromList_noReplace(matchPerms)
if permSeq not in restrictedList:
chosenPerms.append(permSeq)
# Use each chosenSeq to generate the diff matchVersions
for seq in chosenPerms:
# Create Fake miRNA with seq at the seed location to feed to _buildMatchVersions()
seq = 'N%sNNNNNNNNNNNNN' % (bioDefs.revComp(seq))
matchVersions = self._buildMatchVersions(seq)
for seedType in self.matchVersions:
self.matchVersions[seedType][1].append(matchVersions[seedType][0]) # must use index[0] bc _buildMatchVersions returns a list len==1
