def __init__(self, fileName, pseudocountProb=0.0):
self.fileName = fileName
fileHandle = util.get_file_handle(fileName)
self.pseudocountProb = pseudocountProb
self.loadedMotifs = OrderedDict()
action = self.getReadPwmAction(self.loadedMotifs)
util.perform_action_on_each_line_of_file(
file_handle=fileHandle,
action=action,
transformation=util.trim_newline)
for pwm in self.loadedMotifs.values():
pwm.finalise(pseudocountProb=self.pseudocountProb)
super(AbstractLoadedMotifsFromFile, self).__init__(self.loadedMotifs)
def generateSequences(self):
fileHandle = util.get_file_handle(self.dnaseSimulationFile)
for lineNumber, line in enumerate(fileHandle):
if (lineNumber > 0): #ignore title
inp = util.default_tab_seppd(line)
sequenceName = inp[0]
backgroundGenerator = ShuffledBackgroundGenerator(
string=inp[1], shuffler=self.shuffler)
embedders = [
parseDnaseMotifEmbedderString(embedderString,
self.loadedMotifs)
for embedderString in inp[2].split(",")
if len(embedderString) > 0
]
yield SingleDnaseSequenceGenerator(
backgroundGenerator=backgroundGenerator,
dnaseMotifEmbedders=embedders,
sequenceName=sequenceName).generateSequence()
def read_simdata_file(simdata_file, one_hot_encode=False, ids_to_load=None):
ids = []
sequences = []
embeddings = []
labels = []
if (ids_to_load is not None):
ids_to_load = set(ids_to_load)
def action(inp, line_number):
if (line_number > 1):
if (ids_to_load is None or (inp[0] in ids_to_load)):
ids.append(inp[0])
sequences.append(inp[1])
embeddings.append(getEmbeddingsFromString(inp[2]))
labels.append([int(x) for x in inp[3:]])
util.perform_action_on_each_line_of_file(
file_handle=util.get_file_handle(simdata_file),
action=action,
transformation=util.default_tab_seppd)
return util.enum(
ids=ids,
sequences=sequences,
embeddings=embeddings,
labels=np.array(labels))
def printSequences(outputFileName, sequenceSetGenerator,
includeEmbeddings=False, labelGenerator=None,
includeFasta=False, prefix=None):
"""Print a series of synthetic sequences.
Given an output filename, and an instance of
:class:`.AbstractSequenceSetGenerator`, will call the
sequenceSetGenerator and print the generated sequences
to the output file. Will also create a file "info_outputFileName.txt"
in the same directory as outputFileName that contains
all the information about sequenceSetGenerator.
Arguments:
outputFileName: string
sequenceSetGenerator: instance of
:class:`.AbstractSequenceSetGenerator`
includeEmbeddings: a boolean indicating whether to print a
column that lists the embeddings
labelGenerator: optional instance of :class:`.LabelGenerator`
includeFasta: optional boolean indicating whether to also
print out the generated sequences in fasta format
(the file will be produced with a .fa extension)
prefix: string - this will be prefixed in front of the generated
sequence ids, followed by a hyphen
"""
ofh = util.get_file_handle(outputFileName, 'w')
if (includeFasta):
fastaOfh = util.get_file_handle(util.get_file_name_parts(
outputFileName).get_transformed_file_path(
lambda x: x, extension=".fa"), 'w')
ofh.write("seqName\tsequence"
+ ("\tembeddings" if includeEmbeddings else "")
+ ("\t" +
"\t".join(labelGenerator.labelNames)
if labelGenerator is not None else "") + "\n")
generatedSequences = sequenceSetGenerator.generateSequences() # returns a generator
for generatedSequence in generatedSequences:
ofh.write((prefix+"-" if prefix is not None else "")
+ generatedSequence.seqName + "\t" + generatedSequence.seq
+ ("\t" + ",".join(str(x)
for x in generatedSequence.embeddings)
if includeEmbeddings else "")
+ ("\t" + "\t".join(str(x) for x in labelGenerator.generateLabels(
generatedSequence)) if labelGenerator is not None else "")
+ "\n")
if (includeFasta):
fastaOfh.write(">" + (prefix+"-" if prefix is not None else "")
+ generatedSequence.seqName + "\n")
fastaOfh.write(generatedSequence.seq + "\n")
ofh.close()
if (includeFasta):
fastaOfh.close()
infoFilePath = (util.get_file_name_parts(outputFileName)
.get_transformed_file_path(
lambda x: x + "_info", extension=".txt"))
ofh = util.get_file_handle(infoFilePath, 'w')
ofh.write(util.format_as_json(sequenceSetGenerator.getJsonableObject()))
ofh.close()
