def __initParse(self, row):
self.name = row[0]
self.chrom = row[1]
self.strand = row[2]
self.strandRel = False # are these strand-relative coordinates
self.txStart = int(row[3])
self.txEnd = int(row[4])
self.cdsStart = int(row[5])
self.cdsEnd = int(row[6])
exonStarts = intArraySplit(row[8])
exonEnds = intArraySplit(row[9])
iCol = 10
numCols = len(row)
self.score = None
if iCol < numCols: # 10
self.score = int(row[iCol])
iCol = iCol+1
self.name2 = None
if iCol < numCols: # 11
self.name2 = row[iCol]
iCol = iCol+1
self.cdsStartStat = None
self.cdsEndStat = None
if iCol < numCols: # 12,13
self.cdsStartStat = CdsStat(row[iCol])
self.cdsEndStat = CdsStat(row[iCol+1])
iCol = iCol+2
exonFrames = None
self.hasExonFrames = False
if iCol < numCols: # 14
exonFrames = intArraySplit(row[iCol])
iCol = iCol+1
self.hasExonFrames = True
self.__buildExons(exonStarts, exonEnds, exonFrames)
def _initParse(self, row):
self.name = row[0]
self.chrom = row[1]
self.strand = row[2]
self.strandRel = False # are these strand-relative coordinates
self.txStart = int(row[3])
self.txEnd = int(row[4])
self.cdsStart = int(row[5])
self.cdsEnd = int(row[6])
exonStarts = intArraySplit(row[8])
exonEnds = intArraySplit(row[9])
iCol = 10
numCols = len(row)
self.score = None
if iCol < numCols: # 10
self.score = int(row[iCol])
iCol = iCol + 1
self.name2 = None
if iCol < numCols: # 11
self.name2 = row[iCol]
iCol = iCol + 1
self.cdsStartStat = None
self.cdsEndStat = None
if iCol < numCols: # 12,13
self.cdsStartStat = CdsStat(row[iCol])
self.cdsEndStat = CdsStat(row[iCol + 1])
iCol = iCol + 2
exonFrames = None
self.hasExonFrames = False
if iCol < numCols: # 14
exonFrames = intArraySplit(row[iCol])
iCol = iCol + 1
self.hasExonFrames = True
self._buildExons(exonStarts, exonEnds, exonFrames)
def __init__(self, row):
self.numCols = len(row)
self.chrom = row[0]
self.chromStart = int(row[1])
self.chromEnd = int(row[2])
self.name = row[3]
if self.numCols > 4:
self.score = int(row[4])
else:
self.score = None
if self.numCols > 5:
self.strand = row[5]
else:
self.strand = None
if self.numCols > 7:
self.thickStart = int(row[6])
self.thickEnd = int(row[7])
else:
self.thickStart = None
self.thickEnd = None
if self.numCols > 8:
self.itemRgb = row[8]
else:
self.itemRgb = None
if self.numCols > 11:
sizes = intArraySplit(row[10])
relStarts = intArraySplit(row[11])
self.blocks = []
for i in xrange(len(relStarts)):
self.blocks.append(Bed.Block(relStarts[i], sizes[i], self.chromStart+relStarts[i]))
else:
self.blocks = None
def _parseBlockColumns(chromStart, row):
sizes = intArraySplit(row[10])
relStarts = intArraySplit(row[11])
blocks = []
for i in range(len(relStarts)):
start = chromStart + relStarts[i]
blocks.append(Bed.Block(start, start + sizes[i]))
return blocks
def __parseBlocks(self, blockSizesStr, qStartsStr, tStartsStr, qSeqsStr, tSeqsStr):
"convert parallel arrays to PslBlock objects"
self.blocks = []
blockSizes = intArraySplit(blockSizesStr)
qStarts = intArraySplit(qStartsStr)
tStarts = intArraySplit(tStartsStr)
haveSeqs = (qSeqsStr is not None)
if haveSeqs:
qSeqs = strArraySplit(qSeqsStr)
tSeqs = strArraySplit(tSeqsStr)
for i in xrange(self.blockCount):
self.blocks.append(PslBlock(self, qStarts[i], tStarts[i], blockSizes[i],
(qSeqs[i] if haveSeqs else None),
(tSeqs[i] if haveSeqs else None)))
def _parseBlocks(cls, psl, blockCount, blockSizesStr, qStartsStr,
tStartsStr, qSeqsStr, tSeqsStr):
"convert parallel arrays to PslBlock objects"
blockSizes = intArraySplit(blockSizesStr)
qStarts = intArraySplit(qStartsStr)
tStarts = intArraySplit(tStartsStr)
haveSeqs = (qSeqsStr is not None)
if haveSeqs:
qSeqs = strArraySplit(qSeqsStr)
tSeqs = strArraySplit(tSeqsStr)
for i in range(blockCount):
psl.addBlock(
PslBlock(qStarts[i], tStarts[i], blockSizes[i],
(qSeqs[i] if haveSeqs else None),
(tSeqs[i] if haveSeqs else None)))
def _initDb(self, row, dbColIdxMap):
self.name = row[dbColIdxMap["name"]]
self.chrom = row[dbColIdxMap["chrom"]]
self.strand = row[dbColIdxMap["strand"]]
self.strandRel = False # are these strand-relative coordinates
self.txStart = int(row[dbColIdxMap["txStart"]])
self.txEnd = int(row[dbColIdxMap["txEnd"]])
self.cdsStart = int(row[dbColIdxMap["cdsStart"]])
self.cdsEnd = int(row[dbColIdxMap["cdsEnd"]])
exonStarts = intArraySplit(row[dbColIdxMap["exonStarts"]])
exonEnds = intArraySplit(row[dbColIdxMap["exonEnds"]])
self.score = self._colOrNone(row, dbColIdxMap, "score", int)
self.name2 = self._colOrNone(row, dbColIdxMap, "name2", str)
self.cdsStartStat = self._colOrNone(row, dbColIdxMap, "cdsStartStat", CdsStat)
self.cdsEndStat = self._colOrNone(row, dbColIdxMap, "cdsEndStat", CdsStat)
exonFrames = self._colOrNone(row, dbColIdxMap, "exonFrames", intArraySplit)
self.hasExonFrames = (exonFrames is not None)
self._buildExons(exonStarts, exonEnds, exonFrames)
def __initDb(self, row, dbColIdxMap):
self.name = row[dbColIdxMap["name"]]
self.chrom = row[dbColIdxMap["chrom"]]
self.strand = row[dbColIdxMap["strand"]]
self.strandRel = False # are these strand-relative coordinates
self.txStart = int(row[dbColIdxMap["txStart"]])
self.txEnd = int(row[dbColIdxMap["txEnd"]])
self.cdsStart = int(row[dbColIdxMap["cdsStart"]])
self.cdsEnd = int(row[dbColIdxMap["cdsEnd"]])
exonStarts = intArraySplit(row[dbColIdxMap["exonStarts"]])
exonEnds = intArraySplit(row[dbColIdxMap["exonEnds"]])
self.score = self.__colOrNone(row, dbColIdxMap, "score", int)
self.name2 = self.__colOrNone(row, dbColIdxMap, "name2", str)
self.cdsStartStat = self.__colOrNone(row, dbColIdxMap, "cdsStartStat", CdsStat)
self.cdsEndStat = self.__colOrNone(row, dbColIdxMap, "cdsEndStat", CdsStat)
exonFrames = self.__colOrNone(row, dbColIdxMap, "exonFrames", intArraySplit)
self.hasExonFrames = (exonFrames is not None)
self.__buildExons(exonStarts, exonEnds, exonFrames)