def getStatistics( statistics_node, comm_struct):
for son_node in statistics_node:
if son_node.tag.lower() == BedSeqAlignmentStatsCommStruct.MOTIF_STATS_TAG:
name = CommStruct.getAttribute( son_node, BedSeqAlignmentStatsCommStruct.MOTIF_STATS_NAME_ATT)
if name != None and len( name) > 0:
motif_id = CommStruct.getAttribute( son_node, BedSeqAlignmentStatsCommStruct.MOTIF_STATS_ID_ATT)
family = CommStruct.getAttribute( son_node, BedSeqAlignmentStatsCommStruct.MOTIF_STATS_FAMILY_ATT)
classe = CommStruct.getAttribute( son_node, BedSeqAlignmentStatsCommStruct.MOTIF_STATS_CLASS_ATT)
motif_type = CommStruct.getAttribute( son_node, BedSeqAlignmentStatsCommStruct.MOTIF_STATS_TYPE_ATT)
size = CommStruct.getAttributeAsint( son_node, BedSeqAlignmentStatsCommStruct.MOTIF_STATS_SIZE_ATT)
motif_stats = MotifStatistics( name)
motif_stats.motifID = motif_id
motif_stats.motifFamily = family
motif_stats.motifClass = classe
motif_stats.motifType = motif_type
motif_stats.motifSize = size
for param_node in son_node:
if param_node.tag.lower() == BedSeqAlignmentStatsCommStruct.PARAM_TAG:
att_name = CommStruct.getAttribute( param_node, BedSeqAlignmentStatsCommStruct.PARAM_NAME_ATT)
att_value = CommStruct.getAttribute( param_node, BedSeqAlignmentStatsCommStruct.PARAM_VALUE_ATT)
motif_stats.setAttribute( att_name, att_value)
comm_struct.motifStatistics[ name] = motif_stats
elif son_node.tag.lower() == BedSeqAlignmentStatsCommStruct.PARAM_TAG:
att_name = CommStruct.getAttribute( son_node, BedSeqAlignmentStatsCommStruct.PARAM_NAME_ATT)
att_value = CommStruct.getAttribute( son_node, BedSeqAlignmentStatsCommStruct.PARAM_VALUE_ATT)
comm_struct.paramStatistics[ att_name] = att_value
else:
raise ParsingException( "BedSeqAlignmentStatsCommStruct.getStatistics : The statistics contains an unauthorized element : '" + son_node.tag.lower() + "'")
def getAlignmentSequences(sub_node, bedseq, seqalign):
for node_sequence in sub_node:
if node_sequence.tag.lower(
) == BedSeqAlignmentStatsCommStruct.SEQUENCE_TAG:
species = CommStruct.getAttribute(
node_sequence,
BedSeqAlignmentStatsCommStruct.SEQUENCE_SPECIES_ATT)
text = list(
CommStruct.getAttribute(
node_sequence,
BedSeqAlignmentStatsCommStruct.SEQUENCE_TEXT_ATT))
if species != None and text != None:
seqalign.addSequence(species, text)
else:
raise ParsingException(
"BedSeqAlignmentStatsCommStruct.getAlignmentSequences : A sequence of the alignment of '"
+ bedseq.toString() +
"' is missing required attributes")
else:
raise ParsingException(
"BedSeqAlignmentStatsCommStruct.getAlignmentSequences : The sequences of the alignment of '"
+ bedseq.toString() +
"' contains an unauthorized element : '" +
node_sequence.tag.lower() + "'")
def __init__(self):
CommStruct.__init__(self)
self.baseSpecies = ""
# dicitonnary of the BED Sequences : key = <species>.<chromosom> / value = list of BEDSequence
self.bedSequencesDict = {}
# dictionnary of statistics parameters: key = parameter name / Value = parameter value
self.paramStatistics = {}
def __init__(self):
CommStruct.__init__(self)
self.baseSpecies = ""
# dicitonnary of the BED Sequences : key = chromosom / value = list of BED Sequence
self.bedSequencesDict = {}
# dicitonnary of the Multiple Alignments : key = BED Sequence / value = list of multiple alignments
self.bedToMA = {}
def getAlignmentMotifs( sub_node, bedseq, seqalign):
for node_motif in sub_node:
if node_motif.tag.lower() == BedSeqAlignmentStatsCommStruct.MOTIF_TAG:
start = CommStruct.getAttributeAsint( node_motif, BedSeqAlignmentStatsCommStruct.MOTIF_START_ATT)
end = CommStruct.getAttributeAsint( node_motif, BedSeqAlignmentStatsCommStruct.MOTIF_END_ATT)
name = CommStruct.getAttribute( node_motif, BedSeqAlignmentStatsCommStruct.MOTIF_NAME_ATT)
motif_id = CommStruct.getAttribute( node_motif, BedSeqAlignmentStatsCommStruct.MOTIF_ID_ATT, False)
consensus = CommStruct.getAttribute( node_motif, BedSeqAlignmentStatsCommStruct.MOTIF_CONSENSUS_ATT, False)
nb_species = CommStruct.getAttributeAsint( node_motif, BedSeqAlignmentStatsCommStruct.MOTIF_NBSPECIES_ATT, False)
strand = CommStruct.getAttribute( node_motif, BedSeqAlignmentStatsCommStruct.MOTIF_STRAND_ATT, False)
offset = CommStruct.getAttributeAsint( node_motif, BedSeqAlignmentStatsCommStruct.MOTIF_OFFSET_ATT)
score = CommStruct.getAttributeAsfloat( node_motif, BedSeqAlignmentStatsCommStruct.MOTIF_SCORE_ATT)
# Retrieve the PWM of the motif
pwm_s1 = CommStruct.getAttribute( node_motif, BedSeqAlignmentStatsCommStruct.MOTIF_PWM_ATT)
if pwm_s1 != None and len( pwm_s1) > 0:
pwm_matrix = {}
pwm_s2 = pwm_s1.split(";")
for line in pwm_s2:
pwm_s3 = line.split(":")
if len( pwm_s3) > 1:
pwm_s4 = pwm_s3[1].split()
try:
length = 0
for value in pwm_s4:
length += 1
if not pwm_matrix.has_key( pwm_s3[0]):
pwm_matrix[ pwm_s3[0]]=[]
pwm_matrix[ pwm_s3[0]].append( int( value))
except ValueError, val_exce:
raise ParsingException( "BedSeqAlignmentStatsCommStruct.getAlignmentMotifs : Unable to get integer value for Motif '" + BedSeqAlignmentStatsCommStruct.MOTIF_PWM_ATT + "' attributes. From:\n\t---> " + str( val_exce))
pwm = PWM()
pwm.matrix = pwm_matrix
pwm.totalLength = length
pwm.nbSequences = nb_species
else:
pwm = None
if start != None and end != None and name != None:
motif = Motif( start, end, name, pwm)
motif.offset = offset
motif.score = score
if consensus != None:
motif.consensus = consensus
if motif_id != None:
motif.id = motif_id
if strand != None:
motif.strand = strand
seqalign.addMotif( motif)
else:
raise ParsingException( "BedSeqAlignmentStatsCommStruct.getAlignmentMotifs : The motifs of the alignment of '" + bedseq.toString() + "' is missing required attributes")
else:
raise ParsingException( "BedSeqAlignmentStatsCommStruct.getAlignmentMotifs : The motifs of the alignment of '" + bedseq.toString() + "' contains an unauthorized element : '" + node_motif.tag.lower() + "'")
def __init__(self):
CommStruct.__init__(self)
self.baseSpecies = ""
# dicitonnary of the BED Sequences : key = chromosom / value = list of BED Sequence
self.bedSequencesDict = {}
# dictionnary of the Multiple Alignments : key = BED Sequence / value = list of multiple alignments
self.bedToMA = {}
# dictionnary of statistics parameters: key = parameter name / Value = parameter value
self.paramStatistics = {}
# dictionnary of identified motifs statistics : key = motif name / Value = MotifStatistics instance
self.motifStatistics = {}
def getStatistics(statistics_node, comm_struct):
for son_node in statistics_node:
if son_node.tag.lower() == BedSeqCommStruct.PARAM_TAG:
att_name = CommStruct.getAttribute(
son_node, BedSeqCommStruct.PARAM_NAME_ATT)
att_value = CommStruct.getAttribute(
son_node, BedSeqCommStruct.PARAM_VALUE_ATT)
comm_struct.paramStatistics[att_name] = att_value
else:
raise ParsingException(
"BedSeqCommStruct.getStatistics : The statistics contains an unauthorized element : '"
+ son_node.tag.lower() + "'")
def getBEDSequence(node_bedseq, comm_struct):
species = CommStruct.getAttribute(node_bedseq,
BedSeqCommStruct.BEDSEQ_SPECIES_ATT)
chrom = CommStruct.getAttribute(node_bedseq,
BedSeqCommStruct.BEDSEQ_CHROM_ATT)
start = CommStruct.getAttributeAsint(node_bedseq,
BedSeqCommStruct.BEDSEQ_START_ATT)
end = CommStruct.getAttributeAsint(node_bedseq,
BedSeqCommStruct.BEDSEQ_END_ATT)
score = CommStruct.getAttributeAsint(node_bedseq,
BedSeqCommStruct.BEDSEQ_SCORE_ATT,
False)
max = CommStruct.getAttributeAsint(
node_bedseq, BedSeqCommStruct.BEDSEQ_MAX_PEAK_ATT, False)
id = CommStruct.getAttribute(node_bedseq,
BedSeqCommStruct.BEDSEQ_ID_ATT, False)
if species != None and chrom != None and start != None and end != None:
bed_sequence = BEDSequence(species, chrom, start, end)
if score != None:
bed_sequence.score = score
if max != None:
bed_sequence.referenceIndex = max
if id != None:
bed_sequence.id = id
comm_struct.addBEDSequence(bed_sequence)
return bed_sequence
else:
raise ParsingException(
"BedSeqCommStruct.getBEDSequence : Malformed BED Sequence - unable to retrieve sequence information"
)
def getAlignment( align_node, bedseq, comm_struct):
name = CommStruct.getAttribute( align_node, BedSeqAlignmentStatsCommStruct.ALIGNMENT_NAME_ATT)
seq_align = SequenceAlignment()
seq_align.name = name
seq_align.referenceSpecies = bedseq.species
comm_struct.addSequenceAlignment( bedseq, seq_align)
return seq_align
def getMotif(node_motif, comm_struct):
name = CommStruct.getAttribute(
node_motif, MotifStatisticsCommStruct.MOTIF_NAME_ATT)
consensus = CommStruct.getAttribute(
node_motif, MotifStatisticsCommStruct.MOTIF_CONSENSUS_ATT, False)
if name != None:
motif = Motif(0, 0, name, None)
if consensus != None:
motif.consensus = consensus
comm_struct.addMotif(motif)
statistics = MotifStatisticsCommStruct.getMotifStatistics(
node_motif, motif)
comm_struct.addMotifStatistics(motif, statistics)
else:
raise ParsingException(
"MotifStatisticsCommStruct.getAlignmentMotifs : The motif is missing required attribute 'name'"
)
def getBEDSequence( node_bedseq, comm_struct):
try:
species = CommStruct.getAttribute( node_bedseq, BedSeqAlignmentStatsCommStruct.BEDSEQ_SPECIES_ATT)
chrom = CommStruct.getAttribute( node_bedseq, BedSeqAlignmentStatsCommStruct.BEDSEQ_CHROM_ATT)
start = CommStruct.getAttributeAsint( node_bedseq, BedSeqAlignmentStatsCommStruct.BEDSEQ_START_ATT)
end = CommStruct.getAttributeAsint( node_bedseq, BedSeqAlignmentStatsCommStruct.BEDSEQ_END_ATT)
score = CommStruct.getAttributeAsint( node_bedseq, BedSeqAlignmentStatsCommStruct.BEDSEQ_SCORE_ATT, False)
peak_max = CommStruct.getAttributeAsint( node_bedseq, BedSeqAlignmentStatsCommStruct.BEDSEQ_MAX_PEAK_ATT, False)
peak_id = CommStruct.getAttribute( node_bedseq, BedSeqAlignmentStatsCommStruct.BEDSEQ_ID_ATT, False)
except ParsingException, par_exce:
raise ParsingException ( "BedSeqAlignmentStatsCommStruct.getBEDSequence : Malformed BED Sequence - some attributes are not numbers. From:\n\t---> " + str( par_exce))
def __init__(self):
CommStruct.__init__(self)
self.motifList = []
self.motifToStatistics = {}