def __init__(self, strSample, strRef, options, InstInitFolder):
UserFolderAdmin.__init__(self, strSample, strRef, options,
InstInitFolder.strLogPath)
self.strSample = strSample
self._RemoveTmpBeforStart()
self.MakeSampleFolder() ## inheritance
self.strRef = strRef
self.intCore = options.multicore
self.strGapOpen = options.gap_open
self.strGapExtend = options.gap_extend
self.strTargetWindow = options.target_window
self.strIndelCheckPos = options.indel_check_pos
self.strTargetRefAlt = options.target_ref_alt
self.strBarcodeFile = os.path.join(self.strRefDir, 'Barcode.txt')
self.strReferenceSeqFile = os.path.join(self.strRefDir,
'Reference.txt')
self.strRefFile = os.path.join(self.strRefDir, 'Reference.fa')
self.strPamSeq = options.PAM_seq
self.strPamPos = options.PAM_pos
self.strGuidePos = options.Guide_pos
Helper.MakeFolderIfNot(
'./Output/{user}/{project}/{sample}/Tmp/Alignment'.format(
user=self.strUser,
project=self.strProject,
sample=self.strSample))
def SubtractIndelWithD0(dictD0IndelMerge, dictExpIndel):
"""
dictD0IndelMerge: indel proportion - dictExpIndel: indel proportion
"""
strD0SubResultDir = './Output/{user}/{project}/All_results/D0SubResult'.format(
user=strUserName, project=strProjectName)
Helper.MakeFolderIfNot(strD0SubResultDir)
for strSample, dictBarcode in dictExpIndel.items():
with open(
os.path.join(
strD0SubResultDir,
'{sample}_D0SubResult.txt').format(sample=strSample),
'w') as Output:
Output.write(
'Barcode_indel_seq\tD0_total\tD0_indel_prop\tExp_total\tExp_indel_prop\tD0_sub_indel_prop\n'
)
for strBarcode, dictCountTotalAndIndel in dictBarcode.items():
intExpTotal = dictCountTotalAndIndel['Total']
for strIndelSeq, dictCount in dictCountTotalAndIndel.items():
if strIndelSeq == 'Total': continue
try:
intD0Total = dictD0IndelMerge[strBarcode]['Total']
intD0Count = dictD0IndelMerge[strBarcode][strIndelSeq][
'IndelCount']
floD0Prop = round(intD0Count / float(intD0Total), 6)
intExpCount = dictCount['IndelCount']
floExpProp = round(intExpCount / float(intExpTotal), 6)
floSubExpIndel = floExpProp - floD0Prop
if floSubExpIndel < 0:
floSubExpIndel = 0
Output.write('\t'.join(
map(str, [
strIndelSeq, intD0Total, floD0Prop,
intExpTotal, floExpProp, floSubExpIndel
])) + '\n')
except KeyError:
intExpCount = dictCount['IndelCount']
floExpProp = round(intExpCount / float(intExpTotal), 6)
Output.write('\t'.join(
map(str, [
strIndelSeq, 'None', 'None', intExpTotal,
floExpProp, floExpProp
])) + '\n')
def CountGroup(InstParameters):
"""
Sorting_barcode Unique_RandomBarcodeNumber_In_SortingBarcode RandomBarcode Each_RandomBarcode_read_count
TATATCATAGCGTACTCATC 8 TGCGTTTG 3
TATATCATAGCGTACTCATC 8 CGCGTTTG 3
TATATCATAGCGTACTCATC 8 TAGTTTTG 1
TATATCATAGCGTACTCATC 8 ATAGTTTG 1
"""
sHeader = ''
with open(InstParameters.strSampleList) as Sample: ## tmp input
listSample = Sample.readlines()
setGroup = set([
strRow.replace('\n', '').split('\t')[2].upper()
for strRow in listSample
])
for strGroup in setGroup:
if strGroup == 'CTRL': continue
for strRow in listSample:
if strGroup == strGroupOfSample: ## matched group names -> Sum the counts
listCol = strRow.replace('\n', '').split('\t')
strSample = listCol[0]
strRef = listCol[1]
strGroupOfSample = listCol[2]
strProjectDir = './Output/{user}/{project}'.format(
user=InstParameters.strUser,
project=InstParameters.strProject)
strGroupDir = os.path.join(strProjectDir, 'Group_result')
Helper.MakeFolderIfNot(strGroupDir)
dTotal_RandomBarcode_cnt_in_SortingBarcode = OrderedDict(
) ## ('GECKO_6367_GATCTGCTC', ['GECKO_6367', 'GATCTGCTC', 2, 156, '0.0128']),
## Unique key, only one list.
with open('{project_dir}/{sample}_all_random_barcode.txt'.
format(
project_dir=strProjectDir,
sample=strSample)) as RandomBarcode_SeqFreq:
sHeader = RandomBarcode_SeqFreq.readline()
for sRow in RandomBarcode_SeqFreq:
lCol = sRow.replace('\n', '').split('\t')
sSortingBarcode = lCol[0]
#iTotal_RandomBarcode_cnt_in_SortingBarcode = int(lCol[1])
sSorting_and_Random_barcode_seq = lCol[
0] + '_' + lCol[
2] ## Unique name : Doench2014_1000_CTCTGGGGT
iRandomBarcode_count = int(lCol[3])
lCol[3] = iRandomBarcode_count
try:
_ = dTotal_RandomBarcode_cnt_in_SortingBarcode[
sSorting_and_Random_barcode_seq]
dTotal_RandomBarcode_cnt_in_SortingBarcode[
sSorting_and_Random_barcode_seq][
3] += iRandomBarcode_count
except KeyError:
dTotal_RandomBarcode_cnt_in_SortingBarcode[
sSorting_and_Random_barcode_seq] = lCol ## initial assignment
#END for
dRecal_total_kind_of_RandomBarcode = OrderedDict()
for sSort_Rand_seq in dTotal_RandomBarcode_cnt_in_SortingBarcode: ## sSorting_and_Random_barcode_seq
sSortBarcode = sSort_Rand_seq.split('_')[0]
try:
dRecal_total_kind_of_RandomBarcode[
sSortBarcode].append(
dTotal_RandomBarcode_cnt_in_SortingBarcode[
sSort_Rand_seq])
except KeyError:
dRecal_total_kind_of_RandomBarcode[
sSortBarcode] = [
dTotal_RandomBarcode_cnt_in_SortingBarcode[
sSort_Rand_seq]
]
for sKey, llValue in dRecal_total_kind_of_RandomBarcode.items(
):
## sKey: TATATCATAGCGTACTCATC, llValue : [[TATATCATAGCGTACTCATC, 8, TGCGTTTG, 3],[],[] ...
iKind_of_RandomBarcode = len(
llValue
) ################## why do I make like this ?????
for lValue in llValue:
lValue[
1] = iKind_of_RandomBarcode ## Recal using group total cnt.
llValue = sorted(llValue,
key=lambda x: x[3],
reverse=True)
dRecal_total_kind_of_RandomBarcode[sKey] = llValue
strEachGroup = './Output/Group_result/%s' % strGroup
Helper.MakeFolderIfNot(strEachGroup)
with open(os.path.join(strEachGroup, 'Summary_all_random_barcode_in_group.txt'), 'w') as Sort_Random_cnt,\
open(os.path.join(strEachGroup, 'Summary_Unique_RandomBarcodeNumber_in_group.txt'), 'w') as Uniq_random_cnt:
Sort_Random_cnt.write(sHeader)
Uniq_random_cnt.write(
'Sorting_barcode\tUnique_RandomBarcodeNumber_In_SortingBarcode\n'
)
for sSortBarcode, llCol in dRecal_total_kind_of_RandomBarcode.items(
):
Uniq_random_cnt.write(
'\t'.join(map(str, [sSortBarcode,
len(llCol)])) + '\n')
for lCol in llCol:
Sort_Random_cnt.write(
'\t'.join(map(str, lCol)) + '\n')
def __init__(self, strSample, strRef, options, InstInitFolder):
UserFolderAdmin.__init__(self, strSample, strRef, options, InstInitFolder.strLogPath)
self.MakeSampleFolder()
self.strProjectFile = InstInitFolder.strProjectFile
self.intChunkSize = options.chunk_number
self.strQualCutoff = options.base_quality
self.intInsertionWin = options.insertion_window # Insertion window 0,1,2,3,4
self.intDeletionWin = options.deletion_window # Deletion window 0,1,2,3,4
self.strPamType = options.pam_type # CRISPR type : Cpf1(2 cleavages), Cas9(1 cleavage)
self.strPamPos = options.pam_pos # Barcode target position : Forward (barcode + target), Reverse (target + barcode)
self.strPickle = options.pickle
self.strClassFASTQ = options.class_fastq
self.strSplit = options.split
self.strLogPath = InstInitFolder.strLogPath
self.strBarcodeFile = os.path.join(self.strRefDir, 'Barcode.txt')
self.strReferenceSeqFile = os.path.join(self.strRefDir, 'Reference_sequence.txt')
self.strTargetSeqFile = os.path.join(self.strRefDir, 'Target_region.txt')
self.strRefFile = os.path.join(self.strRefDir, 'Reference.fa')
## The file name required for the user is 'B'arcode.txt but it may be written as 'b'arcode.txt by mistake.
## This part is to fix the situation as mentioned above.
if not os.path.isfile(self.strBarcodeFile):
if os.path.isfile(self.strRefDir + 'barcode.txt'):
self.strBarcodeFile = self.strRefDir + 'barcode.txt'
else:
logging.error('Barcode path is not correct, please make sure the path correctly.')
if not os.path.isfile(self.strReferenceSeqFile):
if os.path.isfile(self.strRefDir + 'reference_sequence.txt'):
self.strReferenceSeqFile = self.strRefDir + 'reference_sequence.txt'
else:
logging.error('Reference path is not correct, please make sure the path correctly.')
if not os.path.isfile(self.strTargetSeqFile):
if os.path.isfile(self.strRefDir + 'target_region.txt'):
self.strTargetSeqFile = self.strRefDir + 'target_region.txt'
else:
logging.error('Target path is not correct, please make sure the path correctly.')
self.strFastqDir = './Input/{user}/FASTQ/{project}'.format(user=self.strUser,
project=self.strProject)
## './Input/JaeWoo/FASTQ/Test_samples/Sample_1'
self.strSampleDir = os.path.join(self.strFastqDir, self.strSample)
self.strFastq_name = ''
for strFile in os.listdir(self.strSampleDir):
if os.path.isfile(self.strSampleDir + '/' + strFile) and strFile.split('.')[-1] == 'fastq':
self.strFastq_name = '.'.join(strFile.split('.')[:-1])
logging.info('File name : %s' % self.strFastq_name)
## './Input/JaeWoo/FASTQ/Test_samples/Sample_1/Fastq_file.fastq'
self.strInputFile = os.path.join(self.strSampleDir, self.strFastq_name+'.fastq')
## './Input/JaeWoo/FASTQ/Test_samples/Sample_1/Fastq_file.txt'
self.strInputList = os.path.join(self.strSampleDir, self.strFastq_name+'.txt')
## './Input/JaeWoo/FASTQ/Test_samples/Sample_1/Split_files'
self.strSplitPath = os.path.join(self.strSampleDir, 'Split_files')
Helper.MakeFolderIfNot(self.strSplitPath)
self.strPair = 'False' # FASTQ pair: True, False
def Convert_Indelsearcher_output(strSampleRefGroup):
listSampleRefGroup = strSampleRefGroup.replace('\n',
'').replace('\r',
'').split('\t')
strSample = listSampleRefGroup[0]
strRef = listSampleRefGroup[1]
print('Processing: %s, %s' % (strSample, strRef))
strBaseEditRefFolder = '../Base_edit_2/Input/{user}/Reference/{project}/{ref}'.format(
user=strUser, project=strProject, ref=strRef)
strBaseEditQueryFolder = '../Base_edit_2/Input/{user}/Query/{project}/{sample}'.format(
user=strUser, project=strProject, sample=strSample)
try:
Helper.MakeFolderIfNot(strBaseEditRefFolder)
Helper.MakeFolderIfNot(strBaseEditQueryFolder)
except OSError as e:
print(e)
pass
## BaseEdit refer format : filename, barcode, reference
ReferenceFile_in_IndelSearcher = open(
'./Input/{user}/Reference/{project}/{ref}/Reference_sequence.txt'.
format(user=strUser, project=strProject, ref=strRef))
BarcodeFile_in_IndelSearcher = open(
'./Input/{user}/Reference/{project}/{ref}/Barcode.txt'.format(
user=strUser, project=strProject, ref=strRef))
BarcodeFile_for_BaseEdit = open(
'../Base_edit_2/Input/{user}/Reference/{project}/{ref}/Barcode.txt'.
format(user=strUser, project=strProject, ref=strRef), 'w')
Reference_for_BaseEdit = open(
'../Base_edit_2/Input/{user}/Reference/{project}/{ref}/Reference.txt'.
format(user=strUser, ref=strRef, project=strProject),
'w') ## conversion target to barcode:refseq
dictBarcodeSeq = {}
for strBarcodeIndelSearcher, strReferenceIndelSearcher in zip(
BarcodeFile_in_IndelSearcher, ReferenceFile_in_IndelSearcher):
strBarcodeIndelSearcher = strBarcodeIndelSearcher.replace('\n',
'').strip()
strReferenceIndelSearcher = strReferenceIndelSearcher.replace(
'\n', '').strip()
dictBarcodeSeq[strBarcodeIndelSearcher] = []
BarcodeFile_for_BaseEdit.write(
strBarcodeIndelSearcher + ':' + strBarcodeIndelSearcher + '\n'
) ## first is filename, second is barcode. BaseEdit barcode format
Reference_for_BaseEdit.write(strBarcodeIndelSearcher + ':' +
strReferenceIndelSearcher + '\n')
ReferenceFile_in_IndelSearcher.close()
BarcodeFile_in_IndelSearcher.close()
Reference_for_BaseEdit.close()
Total_result_file = open(
'./Output/{user}/{project}/{sample}/Tmp/{sample}_Classified_Indel_barcode.fastq'
.format(user=strUser, project=strProject, sample=strSample))
intCheckTotLine = 0
intOneLineMore = 0
for i, strRow in enumerate(Total_result_file): ## for query reads
if intOneLineMore == 1:
intCheckTotLine = 0
intOneLineMore = 0
if i % 4 == 0: ## Classified_Indel_barcode has all total sequence.
strBarcode = strRow.split('Barcode_')[1].split(':')[0]
intCheckTotLine = 1
elif intCheckTotLine == 1:
dictBarcodeSeq[strBarcode].append(strRow)
intOneLineMore = 1
for strBarcode, listSeq in dictBarcodeSeq.items():
with open(
'../Base_edit_2/Input/{user}/Query/{project}/{sample}/{barcode}.txt'
.format(user=strUser,
project=strProject,
sample=strSample,
barcode=strBarcode), 'w') as Output:
Output.write(''.join(listSeq))
Total_result_file.close()