merge_or_rename.py

import sys
import os
import re
import glob


#sys.argv[1] = action
#sys.argv[2] = top level folder
#sys.argv[3] = read
def subdirectories(args):
	return [name for name in os.listdir(args)
            if os.path.isdir(os.path.join(args, name))]



def findKey(input_dict, value): #util for dict
	return {k for k, v in input_dict.items() if v == value}


def rchop(thestring, ending): #util for regex 
  if thestring.endswith(ending):
    return thestring[:-len(ending)]
  return thestring

def newName(oldName): #util for regex
	prefix = '[A-Z]_+'
	suffix = '.+(_[bc])'
	noPool = re.match(prefix,oldName).group(0)
	sampleOnly = re.match(suffix,oldName[len(noPool):])
	return rchop(sampleOnly.group(0),sampleOnly.group(1))



def sampleDict(path): #files in dir are made into dictionary, value is R1 or R2
	direc = path # Get current working directory
	ext = '.gz' # Select your file delimiter
	exp = "R[1|2]"

	regex = re.compile(exp)


	samples = {} # Create an empty dict

	# Select only files with the ext extension

	fastq = [i for i in os.listdir(direc) if os.path.splitext(i)[1] == ext]
	for f in fastq:
		
		if regex.search(f) is not None:
				read = re.findall(exp, f)
				samples.update({f:read[0]})

	return samples

def merge(samples): #merges files
	oldName = samples.keys()[1]

	return "cat " + " ".join(sorted([str(x) for x in findKey(samples,sys.argv[3])])) + " > "  + newName(oldName) + "_" + sys.argv[3]+"_"+sys.argv[1]+".fastq.gz"



def rename(samples):
	oldName = samples.keys()[1]
	return "mv " + " ".join(sorted([str(x) for x in findKey(samples,sys.argv[3])])) + " "  + newName(oldName) + "_" + sys.argv[3]+"_"+sys.argv[1]+".fastq.gz"


def main():

	if sys.argv[1] == "merge":
		for subdir in subdirectories(sys.argv[2]):
			os.system(merge(sampleDict(sys.argv[2]+subdir)))
	if sys.argv[1] == "rename":
		for subdir in subdirectories(sys.argv[2]):
			os.system(rename(sampleDict(sys.argv[2]+subdir)))

if __name__ == '__main__':
    main()