batchquark.snake

import os
from glob import glob

mouse_index="/mnt/scratch1/hirak/quark_data/mouse_index/"
human_index="/mnt/scratch1/hirak/quark_data/sailfish_index/"

quark=os.path.abspath("./build/src/quark")

output_path=os.path.abspath(config['stat'])
srrnames = ["/mnt/scratch1/hirak/quark_data/SRR1265495/", 
            "/mnt/scratch1/hirak/quark_data/SRR1265496/",
            "/mnt/scratch1/hirak/quark_data/SRR037452/",
            "/mnt/scratch1/hirak/quark_data/SRR1294122/",
            "/mnt/scratch1/hirak/quark_data/SRR445718/",
            "/mnt/scratch1/hirak/quark_data/SRR490961/",
            "/mnt/scratch1/hirak/quark_data/SRR635193/",
            "/mnt/scratch1/hirak/quark_data/SRR689233/"]

rule runbatchencode:
    run:
        for srrdir in srrnames:
            inputfiles=glob(os.path.join(srrdir,"*.fastq.gz"))
            shell("echo {} {} >> {}".format(os.path.basename(os.path.normpath(srrdir)),len(inputfiles),output_path))
            quark_out=os.path.join(srrdir,"only_quark_out")
            if(not os.path.isdir(quark_out)):
                shell("mkdir -p {}".format(quark_out))
            if(len(inputfiles) == 2):
                mate_1=""
                mate_2=""
                if(inputfiles[0].split("_")[-1].split(".")[0] == "1"):
                    mate_1 = inputfiles[0]
                    mate_2 = inputfiles[1]
                else:
                    mate_2 = inputfiles[0]
                    mate_1 = inputfiles[1]
                #run quark
                #if mouse
                if(os.path.basename(os.path.normpath(srrdir)) == "SRR689233"):
                    shell("/usr/bin/time -o {} -a {} quant -i {} -l IU -1 <(gunzip -c {}) -2 <(gunzip -c {}) -p 20 -o {}"\
                    .format(output_path, quark, mouse_index, mate_1,mate_2,quark_out))
                else:
                    shell("/usr/bin/time -o {} -a {} quant -i {} -l IU -1 <(gunzip -c {}) -2 <(gunzip -c {}) -p 20 -o {}"\
                    .format(output_path, quark, human_index, mate_1,mate_2,quark_out))
            else:
                mate=inputfiles[0]
                shell("/usr/bin/time -o {} -a {} quant -i {} -l U -r <(gunzip -c {}) -p 20 -o {}"\
                    .format(output_path, quark, human_index, mate, quark_out))
rule runbatchdecode:
    run:
        for srrdir in srrnames:
            leonfiles=glob(srrdir,"*.fastq.leon")
            shell("echo {} {} >> {}".format(os.path.basename(os.path.normpath(srrdir)),len(leonfiles),output_path))
            #run leon