tRNAsynthetase.py

import re

class tRNAsynthetase(object):
    def __init__(self, filepath="/Users/coela/projects/ATS_interaction/data/uniprot/aatrnasy.txt"):
        fileHandle = open(filepath,'r')
        match_tRNA = re.compile("(\S+)-tRNA synthetases \((EC \S+)\)")
        count = 0
        switch = False
        self.info = dict()
        self.aadict = {"Alanyl" : "Ala",
                "Cysteinyl" : "Cys",
                "Aspartyl" : "Asp",
                "Glutamyl" : "Glu",
                "Phenylalanyl" : "Phe",
                "Glycyl" : "Gly",
                "Histidyl" : "His",
                "Isoleucyl" : "Ile",
                "Lysyl" : "Lys",
                "Leucyl" : "Leu",
                "Methionyl" : "Met",
                "Asparaginyl" : "Asn",
                "Prolyl" : "Pro",
                "Glutaminyl" : "Gln",
                "Arginyl" : "Arg",
                "Seryl" : "Ser",
                "Threonyl" : "Thr",
                "Valyl" : "Val",
                "Tryptophanyl" : "Trp",
                "Tyrosyl" : "Tyr"}


        for line in fileHandle:
            line = line.rstrip()

            tRNA_data = match_tRNA.match(line)

            if tRNA_data:
                amino_type = tRNA_data.group(1)
                ec_number = tRNA_data.group(2)
                count = 0
                switch = True
            if switch:
                count = count + 1

            if line[0:5] == "Class":
                tRNA_class = line[7:9]
            if line[0:8] == "3D known":
                known = line[10:13]
            if count >= 5 and len(line) == 0:
                count = 0
                continue

            if count >= 5:
                matchlist = re.finditer(r"(\S+)\s+\((\S+)\)",line)
                for match in matchlist:
                    self.info[match.group(1)] = {"aa" : self.aadict[amino_type], "ec" : ec_number, "class" : tRNA_class}
  if __name__ == "__main__":
    tRNAsynthetase = tRNAsynthetase()
    print tRNAsynthetase.info