def generateQuantity(self):
"""See superclass.
"""
val = random.random()
if (val < self.zeroProb):
return 0
else:
return self.quantityGenerator.generateQuantity()
def _embed(self, backgroundStringArr, priorEmbeddedThings, additionalInfo):
"""See superclass.
"""
if (random.random() < self.probOfFirst):
embedder = self.embedder1
else:
embedder = self.embedder2
return embedder.embed(backgroundStringArr, priorEmbeddedThings, additionalInfo)
def sampleIndexWithinRegionOfLength(length, lengthOfThingToEmbed):
"""Uniformly at random samples integers from 0 to
``length``-``lengthOfThingToEmbedIn``.
Arguments:
length: length of full region that could be embedded in
lengthOfThingToEmbed: length of thing being embedded in larger region
"""
assert lengthOfThingToEmbed <= length
indexToSample = int(random.random() *
((length - lengthOfThingToEmbed) + 1))
return indexToSample
def sampleFromProbsArr(arrWithProbs):
"""Samples from a discrete distribution.
Arguments:
arrWithProbs: array of probabilities
Returns:
an index, sampled with the probability of that index in
array of probabilities.
"""
randNum = random.random()
cdfSoFar = 0
for (idx, prob) in enumerate(arrWithProbs):
cdfSoFar += prob
if (cdfSoFar >= randNum or idx == (len(arrWithProbs) - 1)): # need the
# letterIdx==(len(row)-1) clause because of potential floating point errors
# that mean arrWithProbs doesn't sum to 1
return idx
def _generatePos(self, lenBackground, lenSubstring, additionalInfo):
# choose whether to embed in the left or the right
if random.random() > 0.5:
left = True
else:
left = False
# embeddableLength is the length of the region we are considering
# embedding in
embeddableLength = 0.5 * (lenBackground - self.centralBp)
# if lenBackground-self.centralBp is odd, the longer region
# goes on the left (inverse of the shorter embeddable region going on the left in
# the centralBpToEmbedIn case
if (left):
embeddableLength = math.ceil(embeddableLength)
startIndexForRegionToEmbedIn = 0
else:
embeddableLength = math.floor(embeddableLength)
startIndexForRegionToEmbedIn = math.ceil(
(lenBackground - self.centralBp) / 2) + self.centralBp
indexToSample = startIndexForRegionToEmbedIn + \
sampleIndexWithinRegionOfLength(embeddableLength, lenSubstring)
return int(indexToSample)
def _embed(self, backgroundStringArr, priorEmbeddedThings, additionalInfo):
"""Shoves the designated embeddable at the designated position
Skips if some of the positions are already occupied.
"""
embeddable = self.embeddableGenerator.generateEmbeddable()
canEmbed = embeddable.canEmbed(priorEmbeddedThings, self.startPos)
#randomly pick a value for searchLeft
if random.random() < 0.5:
searchLeft = True
else:
searchLeft = False
validEmbeddingPos = self._getValidEmbeddingPos(
embeddable=embeddable,
priorEmbeddedThings=priorEmbeddedThings,
backgroundStringArr=backgroundStringArr,
startingPosToSearchFrom=self.startPos,
searchLeft=searchLeft)
#if couldn't find a valid pos, search in the other direction
if (validEmbeddingPos is None):
validEmbeddingPos = self._getValidEmbeddingPos(
embeddable=embeddable,
priorEmbeddedThings=priorEmbeddedThings,
backgroundStringArr=backgroundStringArr,
startingPosToSearchFrom=self.startPos,
searchLeft=(searchLeft == False))
if (validEmbeddingPos is None):
print("Warning: could not find a place to embed " +
str(embeddable) + "; bailing")
return
else:
embeddable.embedInBackgroundStringArr(
priorEmbeddedThings=priorEmbeddedThings,
backgroundStringArr=backgroundStringArr,
startPos=validEmbeddingPos)
def sampleWithoutReplacement(arr, numToSample):
arrayCopy = [x for x in arr]
for i in range(numToSample):
randomIndex = int(random.random() * (len(arrayCopy) - i)) + i
swapIndices(arrayCopy, i, randomIndex)
return arrayCopy[0:numToSample]
def generateQuantity(self):
"""See superclass.
"""
# the 1+ makes the max val inclusive
return self.minVal + int(random.random() *
(1 + self.maxVal - self.minVal))
def generateQuantity(self):
"""See superclass.
"""
return self.setOfPossibleValues[int(random.random() *
(len(self.setOfPossibleValues)))]
def generateQuantity(self):
"""See superclass.
"""
return 1 if (random.random() <= self.prob) else 0
def generateSubstring(self):
seq, seqDescription = self.substringGenerator.generateSubstring()
if (random.random() < self.reverseComplementProb):
seq = util.reverseComplement(seq)
seqDescription = "revComp-" + seqDescription
return seq, seqDescription
