Esempi in Python per checkForFasta

Esempio n. 1

File: DataSet245rep3.py Progetto: ablackpz/Simulate-mRNASeq-Reads

import makeRandomTestData, os, random

genes = {}
name = '245rep3'
infileName = 'Gallus_gallus.WASHUC2.63.cdna.all.fa'
filename = 'DataSet' + name
genes = makeRandomTestData.inputFastaSeq(infileName)
errors  = 1

if makeRandomTestData.checkForFasta(filename + 'E1R100G100.fa'):
    x = os.remove(filename + 'E1R100G100.fa')
if makeRandomTestData.checkForFasta(filename + 'E1R100G100RandomReadsPair1.fa'):
    x = os.remove(filename + 'E1R100G100RandomReadsPair1.fa')
if makeRandomTestData.checkForFasta(filename + 'E1R100G100RandomReadsPair2.fa'):
    x = os.remove(filename + 'E1R100G100RandomReadsPair2.fa')
if makeRandomTestData.checkForFasta(filename + 'E1R100G100RandomReads.txt'):
    x = os.remove(filename + 'E1R100G100RandomReads.txt')
if makeRandomTestData.checkForFasta(filename + 'E1R90G100.fa'):
    x = os.remove(filename + 'E1R90G100.fa')
if makeRandomTestData.checkForFasta(filename + 'E1R80G100.fa'):
    x = os.remove(filename + 'E1R80G100.fa')
if makeRandomTestData.checkForFasta(filename + 'E1R70G100.fa'):
    x = os.remove(filename + 'E1R70G100.fa')
if makeRandomTestData.checkForFasta(filename + 'E1R60G100.fa'):
    x = os.remove(filename + 'E1R60G100.fa')
if makeRandomTestData.checkForFasta(filename + 'E1R50G100.fa'):
    x = os.remove(filename + 'E1R50G100.fa')
if makeRandomTestData.checkForFasta(filename + 'E1R40G100.fa'):
    x = os.remove(filename + 'E1R40G100.fa')
if makeRandomTestData.checkForFasta(filename + 'E1R30G100.fa'):
    x = os.remove(filename + 'E1R30G100.fa')

Esempio n. 2

File: DataSet278rep2.py Progetto: ablackpz/Simulate-mRNASeq-Reads

import makeRandomTestData, os, random

genes = {}
name = '278rep2'
infileName = 'Mus_musculus.NCBIM37.65.cdna.all.fa'
filename = 'DataSet' + name
genes = makeRandomTestData.inputFastaSeq(infileName)
errors  = 0

if makeRandomTestData.checkForFasta(filename + 'E1R100G100.fa'):
    x = os.remove(filename + 'E1R100G100.fa')
if makeRandomTestData.checkForFasta(filename + 'E1R100G100RandomReadsPair1.fa'):
    x = os.remove(filename + 'E1R100G100RandomReadsPair1.fa')
if makeRandomTestData.checkForFasta(filename + 'E1R100G100RandomReadsPair2.fa'):
    x = os.remove(filename + 'E1R100G100RandomReadsPair2.fa')
if makeRandomTestData.checkForFasta(filename + 'E1R100G100RandomReads.txt'):
    x = os.remove(filename + 'E1R100G100RandomReads.txt')

count = 0
readLength = 100
insertSize = 400
totalReadLength = readLength * 2 + insertSize
numGenes = len(genes)

for key in genes:
    thisGeneName = key
    if len(genes[thisGeneName]) > totalReadLength:
        complete = makeRandomTestData.outputFastaSeq(filename + 'E1R100G100.fa', thisGeneName, genes[thisGeneName])
        level = 20

        complete = makeRandomTestData.makeRandomlyPlacedPairedEndReads(genes[thisGeneName], thisGeneName, readLength, insertSize, level, filename + 'E1R100G100RandomReadsPair1.fa', filename + 'E1R100G100RandomReadsPair2.fa', filename + 'E1R100G100RandomReads.txt', errors, 0)

Esempio n. 3

File: DataSet183rep3.py Progetto: ablackpz/Simulate-mRNASeq-Reads

import makeRandomTestData, os, random

genes = {}
name = '183rep3'
numTranscripts = 5000
randomTrans = 'Random.' + str(numTranscripts) + '.' + name
isoformProb = 10
min = 100
max = 5000

if makeRandomTestData.checkForFasta(randomTrans + '.fa'):
    x = os.remove(randomTrans + '.fa')
for i in range(0, numTranscripts):
    isoform = random.randint(1, 100)
    if isoform <= isoformProb and len(genes) > 2:
        geneSelected = random.choice(genes.keys())
        geneSelectedParts = geneSelected.split(' ')
        geneSelectedId = ''
        for part in geneSelectedParts:
            geneSelectedId += part
        identifier = 'testgene_' + str(i) + ' Isoform_' + geneSelectedId
        thisGeneName, seq = makeRandomTestData.generateIsoforms(identifier, genes[geneSelected], 'random', 0, 0, 0, min, max)
        genes[thisGeneName] = seq
    else:
        thisGeneName =  'testgene_' + str(i) + ' ' + randomTrans
        genes[thisGeneName] = makeRandomTestData.makeKmerCountData(random.randint(101, 4999), 'default')
    complete = makeRandomTestData.outputFastaSeq(randomTrans + '.fa', thisGeneName, genes[thisGeneName])

infileName = randomTrans + '.fa'
filename = 'DataSet' + name
genes = makeRandomTestData.inputFastaSeq(infileName)

Esempio n. 4

File: DataSet201rep3.py Progetto: ablackpz/Simulate-mRNASeq-Reads

import makeRandomTestData, os, random

genes = {}
name = "201rep3"
infileName = "Mus_musculus.NCBIM37.65.cdna.all.fa"
filename = "DataSet" + name
genes = makeRandomTestData.inputFastaSeq(infileName)
errors = 1

if makeRandomTestData.checkForFasta(filename + "E1R100G100.fa"):
    x = os.remove(filename + "E1R100G100.fa")
if makeRandomTestData.checkForFasta(filename + "E1R100G100RandomReads.fa"):
    x = os.remove(filename + "E1R100G100RandomReads.fa")
if makeRandomTestData.checkForFasta(filename + "E1R100G100RandomReads.txt"):
    x = os.remove(filename + "E1R100G100RandomReads.txt")
if makeRandomTestData.checkForFasta(filename + "E1R90G100.fa"):
    x = os.remove(filename + "E1R90G100.fa")
if makeRandomTestData.checkForFasta(filename + "E1R80G100.fa"):
    x = os.remove(filename + "E1R80G100.fa")
if makeRandomTestData.checkForFasta(filename + "E1R70G100.fa"):
    x = os.remove(filename + "E1R70G100.fa")
if makeRandomTestData.checkForFasta(filename + "E1R60G100.fa"):
    x = os.remove(filename + "E1R60G100.fa")
if makeRandomTestData.checkForFasta(filename + "E1R50G100.fa"):
    x = os.remove(filename + "E1R50G100.fa")

count = 0
readLength = 100
numGenes = len(genes)
r90 = int(numGenes * 0.9)
r80 = int(numGenes * 0.8)