hpa.py

import os
import pandas as pd
import utils


def read_cell_types(config_file):
    '''
    Reads the HPA file containing cell type protein expression profiles
    per tissue.
    :param str config_file: path to the configuration file
    :return: pandas dataframe with the protein expression profiles for each tissue and cell type
    '''
    urls = utils.read_config(filepath=config_file, field='urls')

    if 'hpa_single_cell_tissue_url' in urls:
        filename = utils.download_file(url=urls['hpa_single_cell_tissue_url'], data_dir='data')

    hpa_file = utils.read_zipped_file(filepath=filename)
    data = pd.read_csv(hpa_file, sep='\t', header=0)
    data = data.sort_values(by='pTPM', ascending=False).drop_duplicates(['Gene', 'Tissue', 'Cell type'], keep='first')
    data = data[data['pTPM'] > 0.0]
    
    return data


def map_hpa_data(config_file, hpa_data):
    '''
    Map gene identifiers and filetering only tissues relevant in OrthoHPI 2.0
    :param str config_file: path to the config file
    :param dataframe hpa_data: pandas dataframe with the single cell type data from HPA
    :return: mapped dataframe
    '''
    aliases = utils.parse_string_aliases(config_file, sources=['Ensembl_gene'])
    tissues_mapping = {'heart muscle':'heart', 'small intestine':'intestine', 'rectum':'intestine', 'bronchus':'lung', 'colon':'intestine'}
    hpa_data = hpa_data.replace(tissues_mapping)
    hpa_data['Gene'] = hpa_data['Gene'].map(aliases)
    
    tissues = utils.read_config(filepath=config_file, field='tissues')
    hpa_data = hpa_data[hpa_data['Tissue'].isin([t.lower() for t in tissues.values()])]
    
    return hpa_data


def filter_valid_proteins(hpa_data, valid_proteins):
    hpa_data = hpa_data[hpa_data['Gene'].isin(valid_proteins)]

    return hpa_data


def parse_hpa(config_file, valid_proteins):
    data = read_cell_types(config_file=config_file)
    data = map_hpa_data(config_file=config_file, hpa_data=data)
    data = filter_valid_proteins(data, valid_proteins=valid_proteins)
    
    return data

if __name__ == '__main__':
    config_file='config.yml'
    valid_proteins=['9606.ENSP00000302111', '9606.ENSP00000435412']
    parse_hpa(config_file, valid_proteins)