Factor Parser for GEO

factor_parse.py

def parseFactorByAntibody(geoPost):
    targetFlds = ["CHIP_ANTIBODY", "ANTIBODY", "CHIP", "ANTIBODY_SOURCE", "ANTIBODY_ANTIBODYDESCRIPTION",
                  "ANTIBODY_TARGETDESCRIPTION"]
    #1. try to get the values
    for t in targetFlds:
        tmp = getFromPost(geoPost, t).strip()
        if not tmp:
            continue
        tmp = tmp.upper().replace("ANTI-", " ").replace("ANTI", " ").strip()
        if len(tmp) < 10 and tmp != "":
            return models.Factors.objects.get_or_create(name=tmp)[0]

        splited = re.findall(r"[\w-]+", tmp)
        for s in splited:

            if re.match(r"^[\d-]+$", s):
                continue

            if d.check(s):
                continue

            if s.startswith("POL2") and len(s) < 10:
                return models.Factors.objects.get_or_create(name=tmp)[0]

            if models.Factors.objects.filter(name__iexact=s):
                return models.Factors.objects.get(name__iexact=s)
        if "INPUT" in splited:
            return models.Factors.objects.get_or_create(name="Input")[0]
        if ("POLYMERASE" in splited) or ("POL" in splited):
            return models.Factors.objects.get_or_create(name="POL2")[0]

    return None