def rewrite_align(U, NU, aliDfile, pScoreCutoff, aliFormat, outdir):
pathoUtilsA.ensure_dir(outdir)
f = os.path.basename(aliDfile)
reAlignfile = outdir + os.sep + 'updated_' + f
with open(reAlignfile, 'w') as of:
with open(aliDfile, 'r') as in1:
h_readId = {}
h_refId = {}
genomes = []
read = []
gCnt = 0
rCnt = 0
mxBitSc = 700
sigma2 = 3
for ln in in1:
if (ln[0] == '@' or ln[0] == '#'):
of.write(ln)
continue
l = ln.split('\t')
readId = l[0]
if (aliFormat == 0 or aliFormat == 1): # gnu-sam or sam
#refId=l[2].split("ti:")[-1]
refId = l[2]
if int(
l[1]
) & 0x4 == 4: # bitwise FLAG - 0x4 : segment unmapped
continue
elif (aliFormat == 2): # bl8
refId = l[1]
if refId == '*':
continue
mObj = re.search(r'ti\|(\d+)\|org\|([^|]+)\|', refId)
if mObj:
refId = "ti|" + mObj.group(1) + "|org|" + mObj.group(2)
else:
mObj = re.search(r'ti\|(\d+)\|', refId)
if mObj and mObj.group(1) != "-1":
refId = "ti|" + mObj.group(1)
(_, skipFlag) = find_entry_score(ln, l, aliFormat,
pScoreCutoff)
if skipFlag:
continue
gIdx = h_refId.get(refId, -1)
if gIdx == -1:
gIdx = gCnt
h_refId[refId] = gIdx
genomes.append(refId)
gCnt += 1
rIdx = h_readId.get(readId, -1)
if rIdx == -1:
#hold on this new read
#first, wrap previous read profile and see if any previous read has a same profile with that!
rIdx = rCnt
h_readId[readId] = rIdx
read.append(readId)
rCnt += 1
if rIdx in U:
of.write(ln)
continue
if rIdx in NU:
if (aliFormat == 0): # gnu-sam
scoreComponents = l[12].split(':')
(upPscore,
pscoreSum) = find_updated_score(NU, rIdx, gIdx)
scoreComponents[2] = str(upPscore * pscoreSum)
if (scoreComponents[2] < pScoreCutoff):
continue
l[12] = ':'.join(scoreComponents)
ln = '\t'.join(l)
of.write(ln)
elif (aliFormat == 1): # sam
(upPscore,
pscoreSum) = find_updated_score(NU, rIdx, gIdx)
if (upPscore < pScoreCutoff):
continue
if (upPscore >= 1.0):
upPscore = 0.999999
mapq2 = math.log10(1 - upPscore)
l[4] = str(int(round(-10.0 * mapq2)))
ln = '\t'.join(l)
of.write(ln)
elif (aliFormat == 2): # bl8
(upPscore,
pscoreSum) = find_updated_score(NU, rIdx, gIdx)
score = upPscore * pscoreSum
if score <= 0.0:
continue
bitSc = math.log(score)
if bitSc > mxBitSc:
bitSc = mxBitSc
l[10] = str(bitSc * sigma2)
ln = '\t'.join(l)
of.write(ln)
return reAlignfile
def rewrite_align(U, NU, aliDfile, pScoreCutoff, aliFormat, outdir):
pathoUtilsA.ensure_dir(outdir)
f = os.path.basename(aliDfile)
reAlignfile = outdir + os.sep + 'updated_' + f
with open(reAlignfile,'w') as of:
with open(aliDfile,'r') as in1:
h_readId = {}
h_refId = {}
genomes = []
read =[]
gCnt = 0
rCnt = 0
mxBitSc = 700
sigma2 = 3
for ln in in1:
if (ln[0] == '@' or ln[0] == '#'):
of.write(ln)
continue
l = ln.split('\t')
readId=l[0]
if (aliFormat == 0 or aliFormat == 1): # gnu-sam or sam
#refId=l[2].split("ti:")[-1]
refId=l[2]
if int(l[1])&0x4 == 4: # bitwise FLAG - 0x4 : segment unmapped
continue
elif (aliFormat == 2): # bl8
refId=l[1]
if refId == '*':
continue
mObj=re.search(r'ti\|(\d+)\|org\|([^|]+)\|gi',refId)
if mObj:
refId = "ti|"+mObj.group(1)+"|org|"+mObj.group(2)
else:
mObj=re.search(r'ti\|(\d+)\|gi',refId)
if mObj and mObj.group(1)!="-1":
refId = "ti|"+mObj.group(1)
(_, skipFlag) = find_entry_score(ln, l, aliFormat, pScoreCutoff)
if skipFlag:
continue
gIdx = h_refId.get(refId,-1)
if gIdx == -1:
gIdx = gCnt
h_refId[refId] = gIdx
genomes.append(refId)
gCnt += 1
rIdx = h_readId.get(readId,-1)
if rIdx == -1:
#hold on this new read
#first, wrap previous read profile and see if any previous read has a same profile with that!
rIdx = rCnt
h_readId[readId] = rIdx
read.append(readId)
rCnt += 1
if rIdx in U:
of.write(ln)
continue
if rIdx in NU:
if (aliFormat == 0): # gnu-sam
scoreComponents = l[12].split(':')
(upPscore, pscoreSum) = find_updated_score(NU, rIdx, gIdx)
scoreComponents[2] = str(upPscore*pscoreSum)
if (scoreComponents[2] < pScoreCutoff):
continue
l[12] = ':'.join(scoreComponents)
ln = '\t'.join(l)
of.write(ln)
elif (aliFormat == 1): # sam
(upPscore, pscoreSum) = find_updated_score(NU, rIdx, gIdx)
if (upPscore < pScoreCutoff):
continue
if (upPscore >= 1.0):
upPscore = 0.999999
mapq2 = math.log10(1 - upPscore)
l[4] = str(int(round(-10.0*mapq2)))
ln = '\t'.join(l)
of.write(ln)
elif (aliFormat == 2): # bl8
(upPscore, pscoreSum) = find_updated_score(NU, rIdx, gIdx)
score = upPscore*pscoreSum
if score <= 0.0:
continue
bitSc = math.log(score)
if bitSc > mxBitSc:
bitSc = mxBitSc
l[10] = str(bitSc*sigma2)
ln = '\t'.join(l)
of.write(ln)
return reAlignfile