description=
"TagDigger v. 1.1 barcode splitter command line script by Lindsay V. Clark"
)
parser.add_argument('-b',
'--barcodefile',
help='Name of barcode key file',
required=True)
parser.add_argument('-a',
'--adapter',
help='Name of the adapter set',
required=True,
choices=sorted(tagdigger_fun.adapters.keys()))
args = parser.parse_args()
bckeys = tagdigger_fun.readBarcodeKeyfile(args.barcodefile, forSplitter=True)
if bckeys == None:
raise Exception("Problem reading barcode file.")
adapter = tagdigger_fun.adapters[args.adapter]
enzyme = args.adapter[:args.adapter.find("-")]
cutsite = tagdigger_fun.enzymes[enzyme]
fqfiles = sorted(bckeys.keys())
fqok = [tagdigger_fun.isFastq(f) for f in fqfiles]
if not all(fqok):
print("Cannot read the following as FASTQ files:")
print([fqfiles[i] for i in range(len(fqfiles)) if not fqok[i]])
print("Cut site: " + cutsite)
# set working directory for finding files
tagdigger_fun.set_directory_interactive()
# read in tags
tags = tagdigger_fun.readTags_interactive()
# sanitize
tags = tagdigger_fun.sanitizeTags(tags)
print("{} tag sequences remain.\n".format(len(tags[1])))
# read in key file
bckeys = None
while bckeys == None:
bckeys = tagdigger_fun.readBarcodeKeyfile(
input("Name of key file with barcodes: ").strip())
# summarize
fqfiles = sorted(bckeys.keys())
for f in fqfiles:
print("File {}: {} barcodes".format(f, len(bckeys[f][0])))
print("")
# check that FASTQ files are okay
fqok = [tagdigger_fun.isFastq(f) for f in fqfiles]
while not all(fqok):
print("Cannot read the following as FASTQ files:")
for f in range(len(fqok)):
if not fqok[f]:
print(fqfiles[f])
thischoice = '0'
while thischoice not in {'1', '2', '3'}:
thischoice = input("Use different directory for reading and writing files? (y/n) ").strip()
if thischoice.upper() == 'Y':
dirchoice = ""
while not os.path.isdir(dirchoice):
dirchoice = input("New directory: ")
os.chdir(dirchoice)
print("\nContents of current directory:")
thisdircontents = os.listdir('.')
for i in thisdircontents:
print(i)
# read in key file
bckeys = None
while bckeys == None:
bckeys = tagdigger_fun.readBarcodeKeyfile(input("\nName of key file with barcodes: ").strip(), forSplitter=True)
# summarize
fqfiles = sorted(bckeys.keys())
for f in fqfiles:
print("File {}: {} barcodes".format(f, len(bckeys[f][0])))
print("")
# check that FASTQ files are okay
fqok = [tagdigger_fun.isFastq(f) for f in fqfiles]
while not all(fqok):
print("Cannot read the following as FASTQ files:")
for f in range(len(fqok)):
if not fqok[f]:
print(fqfiles[f])
thischoice = '0'
while thischoice not in {'1', '2', '3'}:
while thischoice not in {"1", "2", "3", "4"}:
thischoice = input("Enter the number of the format of your tag file: ").strip()
tagfile = input("Enter the file name: ").strip()
tags = tagfunctions[thischoice](tagfile, toKeep=toKeep)
print("")
# summarize results
print("{} tag sequences read.\n".format(len(tags[1])))
# sanitize
tags = tagdigger_fun.sanitizeTags(tags)
print("{} tag sequences remain.\n".format(len(tags[1])))
# read in key file
bckeys = None
while bckeys == None:
bckeys = tagdigger_fun.readBarcodeKeyfile(input("Name of key file with barcodes: ").strip())
# summarize
fqfiles = sorted(bckeys.keys())
for f in fqfiles:
print("File {}: {} barcodes".format(f, len(bckeys[f][0])))
print("")
# check that FASTQ files are okay
fqok = [tagdigger_fun.isFastq(f) for f in fqfiles]
while not all(fqok):
print("Cannot read the following as FASTQ files:")
for f in range(len(fqok)):
if not fqok[f]:
print(fqfiles[f])
thischoice = "0"
while thischoice not in {"1", "2", "3"}:
writeMarkerKey = args.TASSELkeyFile != None
tags = tagdigger_fun.readTags_TASSELSAM(args.TASSELSAM,
toKeep=toKeep,
binaryOnly=binaryOnly,
writeMarkerKey=writeMarkerKey,
keyfilename=args.TASSELkeyFile)
if tagformat[6]:
tags = tagdigger_fun.readTags_pyRAD(args.pyRADalleles,
toKeep=toKeep,
binaryOnly=binaryOnly)
if tags == None:
raise Exception("Problem reading tags.")
tags = tagdigger_fun.sanitizeTags(tags)
# read barcode file
bckeys = tagdigger_fun.readBarcodeKeyfile(args.barcodefile)
if bckeys == None:
raise Exception("Problem reading barcode file.")
# check fastq files
fqfiles = sorted(bckeys.keys())
fqok = [tagdigger_fun.isFastq(f) for f in fqfiles]
if not all(fqok):
print("Cannot read the following as FASTQ files:")
print([fqfiles[i] for i in range(len(fqfiles)) if not fqok[i]])
raise Exception("Cannot read all FASTQ files.")
# check binary SNPs for numeric genotype output
if args.outputgen != None:
if set([t[-1] for t in tags[0]]) != {'0', '1'}:
raise Exception(
"Cannot output numeric genotypes for non-binary markers.")
adaptersets = sorted(tagdigger_fun.adapters.keys())
for a in adaptersets:
if enzchoice not in knownenzymes or enzchoice in a:
print(a)
print("")
adaptchoice = ""
while adaptchoice not in adaptersets:
adaptchoice = input("Choose an adapter set: ").strip()
# set working directory for finding files
tagdigger_fun.set_directory_interactive()
# read in key file
bckeys = None
while bckeys == None:
bckeys = tagdigger_fun.readBarcodeKeyfile(
input("\nName of key file with barcodes: ").strip(), forSplitter=True)
# summarize
fqfiles = sorted(bckeys.keys())
for f in fqfiles:
print("File {}: {} barcodes".format(f, len(bckeys[f][0])))
print("")
# check that FASTQ files are okay
fqok = [tagdigger_fun.isFastq(f) for f in fqfiles]
while not all(fqok):
print("Cannot read the following as FASTQ files:")
for f in range(len(fqok)):
if not fqok[f]:
print(fqfiles[f])
thischoice = '0'
while thischoice not in {'1', '2', '3'}:
