XML to TSV parser for Michelle.

parse-michelled-xml.py

import xml.etree.ElementTree
import re

root = xml.etree.ElementTree.parse('RNAi.fixed.xml').getroot()
pattern = re.compile("\w")

def get_text_list(child,path):
    l = list()
    for i in child.findall(path):
        if pattern.match(i.text):
            l.append(str(i.text).rstrip())
    return ';'.join(l)

def get_inhibit_children(child,path):
    l = list()
    for i in child.findall(path):
        for j in i.findall('*'):
            if pattern.match(j.text):
                l.append(str(j.text).rstrip())
                # print('j',str(j.text).rstrip())
                for x in j.findall('**'):
                    l.append(str(x.text).rstrip())
                    # print('x',str(x.text).rstrip())

    return ','.join(l)

print('\t'.join(['RNAi','History_name','Homol','PCR_product','Laboratory','Date','Predicted_gene','Gene','Transcript','Species','Paper','Phenotype']))

for child in root:

    rnai_entry = list()
    rnai_entry.append(str(child.text).rstrip())
    if child.find('History_name') == None:
        rnai_entry.append("NA")
    else:
        rnai_entry.append(child.find('History_name').find('Text').text)
    rnai_entry.append(get_text_list(child,'.//Homol_data'))
    rnai_entry.append(get_text_list(child,'.//PCR_product'))
    if child.find('Experiment') == None:
        rnai_entry.append("NA")
        rnai_entry.append("NA")
    else:
        if child.find('Experiment').find('Laboratory') == None:
            rnai_entry.append("NA")
        else:
            rnai_entry.append(child.find('Experiment').find('Laboratory').find('Laboratory').text)

        if child.find('Experiment').find('Date') == None:
            rnai_entry.append("NA")
        else:
            rnai_entry.append(child.find('Experiment').find('Date').find('Date').text)
    rnai_entry.append(get_inhibit_children(child,'.//Predicted_gene'))
    rnai_entry.append(get_inhibit_children(child,'.//Gene'))
    rnai_entry.append(get_inhibit_children(child,'.//Transcript'))
    if child.find('./Species/Species') == None:
        rnai_entry.append("NA")
    else:
        rnai_entry.append(child.find('./Species/Species').text)

    if child.find('./Reference/Paper') == None:
        rnai_entry.append("NA")
    else:
        rnai_entry.append(child.find('./Reference/Paper').text)

    phen_list = list()

    for i in child.findall('./Phenotype//Phenotype'):
        phen = list()
        phen.append(str(i.text).rstrip())
        for x in i.findall('*'):
            for j in x.findall('*'):
                phen.append(str(j.text).rstrip())
                for y in j.findall('*'):
                    phen.append(str(y.text).rstrip())

        phen_list.append(','.join(phen))

    rnai_entry.append(';'.join(phen_list))
    print('\t'.join(rnai_entry))