SeqSelect.py

import sys

fastaDict = {}


def fastaParser(infile):
    currentGene = ""

    with open(infile) as f:
        for line in f:
            line = line.strip()
            if len(line) > 0 and line[0] == ">":
                currentGene = line[1:].split(" ")[0]
                fastaDict[currentGene] = ""
            else:
                line = line.upper()
                fastaDict[currentGene] += line


def complement(seq):
    ntDict = {"A": "T", "T": "A", "G": "C", "C": "G"}
    tmp = ""

    for nt in range(len(seq)):
        tmp += ntDict[seq[nt]]
    return tmp


def selectReverseSeq(seq, pre, post):
    tmp = ""
    tmp = seq[int(post) - 1:int(pre) - 2:-1].upper()
    return tmp


def runSeqSelect(infile, geneName, pre, post):
    pre = int(pre)
    post = int(post)
    if post > pre:
        tmpSeq = ""
        fastaParser(infile)
        if geneName in fastaDict.keys():
            tmpSeq = fastaDict[geneName]
        return tmpSeq[pre - 1:post]
    else:
        tmpSeq = ""
        fastaParser(infile)
        if geneName in fastaDict.keys():
            tmpSeq = fastaDict[geneName]
        tmpSeq = tmpSeq[pre - 1:post - 2:-1]
        return complement(tmpSeq)


if __name__ == "__main__":
    if len(sys.argv) != 5:
        print
        "python [seqSelect] [fastaFile] [geneName] [startPostion] [endPosition]"
        print
        sys.argv
    else:
        print(runSeqSelect(sys.argv[1], sys.argv[2], sys.argv[3], sys.argv[4]))