comparison SAINT_preprocessing.py @ 51:09b89febcf98 draft

Uploaded

50:61b792d17660

    # Remake inter file without protein errors from Uniprot.
    err = readtab("error proteins.txt")
    bait = readtab(baitfile)
    data = read_Scaffold(Scaffold_input).data
    header = read_Scaffold(Scaffold_input).header
    header = [MQ_var.replace(r"\"", "") for MQ_var in header]
    header = [MQ_var.replace(r"Intensity.", r"") for MQ_var in header]
    header = [MQ_var.replace(r".", r"-") for MQ_var in header]
    bait_index = []
    for bait_item in bait:
        bait_index.append(header.index(bait_item[0]))
    proteins = read_Scaffold(Scaffold_input).proteins
    errors = []
    valid_prots = []
    for e in err:
        errors.append(e[0])
    for a in proteins:
        a = a.replace("\n", "")
        # Remove \n for input into function.
        a = a.replace("\r", "")
        # Ditto for \r.
        seq = get_info(a).seqlength
        GN = get_info(a).genename
        if seq != 'NA':
            if GN != 'NA':
                valid_prots.append(a)
    with open('inter.txt', 'w') as input_file:
        l = 0; a = 0
        for bb in bait:
            for lst in data:
                print lst
                if lst[0] in valid_prots:
                    input_file.write(header[bait_index[l]] + '\t' + bb[1] + '\t' + lst[0] + '\t' + lst[bait_index[l]] + '\n')
                a += 1
                if a == len(proteins):
                    l += 1; a = 0



def bait_check(bait, Scaffold_input): 
    # Check that bait names share Scaffold header titles.
    bait_in = readtab(bait)

51:09b89febcf98

    # Remake inter file without protein errors from Uniprot.
    err = readtab("error proteins.txt")
    bait = readtab(baitfile)
    data = read_Scaffold(Scaffold_input).data
    header = read_Scaffold(Scaffold_input).header
    bait_index = []
    for bait_line in bait:
        bait_index.append(header.index(bait_line[0]))
    proteins = read_Scaffold(Scaffold_input).proteins
    errors = []
    for e in err:
        errors.append(e[0])
    with open('inter.txt', 'w') as y:
        l = 0; a = 0
        for bb in bait:
            for lst in data:
                if proteins[a] not in errors:
                    y.write(header[bait_index[l]] + '\t' + bb[1] + '\t' + proteins[a] + '\t'
                            + lst[bait_index[l]] + '\n')
                a += 1
                if a == len(proteins):
                    l += 1; a = 0


def bait_check(bait, Scaffold_input): 
    # Check that bait names share Scaffold header titles.
    bait_in = readtab(bait)