def _initJASPAR(self,motifData):
"""Sets Motif's attributes using a JASPAR motif record."""
def _buildPWM(dataString):
"""Convert jaspars's matrix data to my format and
return the PWM/barcode.
dataString = 'A [0 3]\nC [94 75]\nG [1 0]\nT [2 19]"""
dataString = dataString.replace('[',' ')
dataString = dataString.replace(']',' ')
data = dataString.split('\n')
data = [x.split() for x in data]
pwm = {}
for d in data:
pwm[d[0]] = tuple([int(x) for x in d[1:]])
bc = hashlib.md5(str(pwm)).hexdigest()
return pwm,bc
acc,mName = motifData[0].lstrip('>').rstrip('\n').split()
self.accession = acc
self.name = mName
self.pwm,self.barcode = _buildPWM(motifData[1])
PWM.__init__(self,self._getMotilityMatrix(self.pwm)) # call motility's PWM's init
self.consensus = self._generateConsensus()
self.source = 'jaspar'
self._setID()
def _initSCOPE(self,motifData):
"""Sets Motif's attributes using a single SCOPE 'motif' data set."""
def _scopeBaseWeights(base,motifData):
"""Return list of base counts/freqs for all columns
of motif."""
baseWeights = []
bases = motifData.getElementsByTagName('base')
for b in bases:
if b.attributes.getNamedItem('name').value.upper().startswith(base):
for w in b.getElementsByTagName('weight'):
#wVal = w.childNodes[0].data
baseWeights.append(float(w.childNodes[0].data))
#break
return tuple(baseWeights)
def _scopePWM(motifData):
"""Returns pwm and barcode data from scope motif."""
pwm = {}
for base in self.bases:
pwm[base] = _scopeBaseWeights(base,motifData)
bc = hashlib.md5(str(pwm)).hexdigest()
return pwm,bc
def _scopeInstances(motifData):
"""Returns tuple of namedtuples:
keys=(sequence, strand<-/+>,start[negVersion],end[negVersion],geneName))"""
# Build instances namedtuples:
Instance = namedtuple('instance','sequence strand begin end gene')
instList = [Instance(x.getAttribute('sequence'),
x.getAttribute('strand'),
x.getAttribute('begin'),
x.getAttribute('end'),
x.getAttribute('gene')) for x in motifData.getElementsByTagName('instance')]
return tuple(instList)
self.consensus = motifData.getAttribute('sequence')
self.sigvalue = motifData.getAttribute('sigvalue')
self.algorithm = motifData.getAttribute('algorithm')
self.genes = tuple(sorted(list(set([x.getAttribute('gene') for x in motifData.getElementsByTagName('instance')])))) # (sorted tuple)
self.instances = _scopeInstances(motifData) # tuple of namedtuples keys=(sequence, strand<-/+>,start[negVersion],end[negVersion],geneName))
self.name = '%s_%s' % (self.sigvalue,self.consensus)
self.accession = '%s_%s' % (self.consensus,int(float(self.sigvalue)))
self.pwm,self.barcode = _scopePWM(motifData) # key(nulceotide),value(tuple of freqs at each position)
self._setID()
PWM.__init__(self,self._getMotilityMatrix(self.pwm))
def _initXMS(self,motifData):
"""Sets Motif's attributes using a single 'motif' node from a
minidom object representing the XMS file."""
def _xmsBaseWeights(base,motifData):
"""Return list of base counts/freqs for all columns
of motif."""
baseWeights = []
weights = motifData.getElementsByTagName('weight')
for w in weights:
if w.attributes.getNamedItem('symbol').value.upper().startswith(base):
baseWeights.append(float(w.childNodes[0].data))
return tuple(baseWeights)
def _xmsProp(motifData,propKey):
"""Returns stored property values if they exist from XMS motif."""
props = motifData.getElementsByTagName('prop')
propVal = None
for p in props:
if p.getElementsByTagName('key')[0].childNodes[0].data.lower() == propKey.lower():
propVal = p.getElementsByTagName('value')[0].childNodes[0].data
break
return propVal
def _xmsPWM(motifData):
"""Returns pwm and barcode data from xms motif."""
pwm = {}
for base in self.bases:
pwm[base] = _xmsBaseWeights(base,motifData)
bc = hashlib.md5(str(pwm)).hexdigest()
return pwm,bc
# init instance attributes from xms motifData
self.name = _xmsProp(motifData,'name')
self.accession = _xmsProp(motifData,'accession')
self.consensus = _xmsProp(motifData,'consensus')
self.sigvalue = _xmsProp(motifData,'sigvalue')
self.rank = _xmsProp(motifData,'rank')
self.algorithm = _xmsProp(motifData,'algorithm')
self.pwm,self.barcode = _xmsPWM(motifData)
self._setID()
PWM.__init__(self,self._getMotilityMatrix(self.pwm))