DB_INPUT_UniPPIN_function

############################################################################
# Populating database with the content of UNIPPIN2018 files
############################################################################

from sqlite3 import connect

DB_name = 'DB_JoRo_PRO'

############################################################################
conn = connect(DB_name)
curs = conn.cursor()

def UNIPPIN_import_to_DB(file_name, table_name):
    """ Function to read .sif files, replace '\t' with ',', and save as .csv
    Converted file is imported into relevant table in the database
    .csv files are kept for QC purposes"""
    
    query = 'INSERT INTO ' + table_name + ' (UniProt_1, UniProt_2) VALUES (?,?)'
    # Replacing '\t' with ',' and saving as .csv
    with open(file_name) as fin, open(file_name + '.csv', 'w') as fout:
        for line in fin:
            fout.write(line.replace('\tppi\t',','))
    file = open(file_name + '.csv')
    rows = [line.rstrip().split(',') for line in file]
    for rec in rows: 
        curs.execute(str(query), rec)
    conn.commit()




UNIPPIN_import_to_DB('BioGRID_noSelfLoop.sif', 'UP_BioGRID_noSelfLoop')
UNIPPIN_import_to_DB('BioPlex_noSelfLoop.sif', 'UP_BioPlex_noSelfLoop')
UNIPPIN_import_to_DB('BPMS_noSelfLoop.sif', 'UP_BPMS_noSelfLoop')
UNIPPIN_import_to_DB('DIPPPI_noSelfLoop.sif', 'UP_DIPPPI_noSelfLoop')
UNIPPIN_import_to_DB('HeinMS_noSelfLoop.sif', 'UP_HeinMS_noSelfLoop')
UNIPPIN_import_to_DB('HPRD_noSelfLoop.sif', 'UP_HPRD_noSelfLoop')
UNIPPIN_import_to_DB('IntAct_noSelfLoop.sif', 'UP_IntAct_noSelfLoop')
UNIPPIN_import_to_DB('Metazoa_noSelfLoop.sif', 'UP_Metazoa_noSelfLoop')
UNIPPIN_import_to_DB('String_noSelfLoop.sif','UP_String_noSelfLoop')
UNIPPIN_import_to_DB('UniPPIN_ACC_noSelfLoop.sif','UP_UniPPIN_ACC_noSelfLoop')
UNIPPIN_import_to_DB('Y2H_noSelfLoop.sif', 'UP_Y2H_noSelfLoop')