#! /usr/bin/env python

"""
# -----------------------------------------------------------------------
# encoding: utf-8

# parse_annotations.py
# Kevin Specht
# Last Modified: 2017-04-27

# This code takes a network and a number and outputs a set of that
#  number of random genes
# -----------------------------------------------------------------------
"""

# -----------------------------------------------------
    # Checking for input from the command line:
    # -----------------------------------------------------
    #
    # [1] file with the seed genes (if table contains more than one
    #     column they must be tab-separated; the first column will be
    #     used only)
    #
    # [2] file providing the network in the form of an edgelist
    #     (tab-separated table, columns 1 & 2 will be used)
    #
    # [3] number of desired iterations
    #
    # [4] name for the results file

import sys
import time
import copy
import csv
import random

"""
Gets the set of seed genes
"""
def get_seed_genes():
    f = open(sys.argv[1],'r')
    seedList=[]
    for line in f:
        seedList.append(line.rstrip())
    f.close()
    return seedList

"""
Gets the links from the network
"""
def get_links(list):
    f=open(list,'r')
    linkList=[]
    for line in f:
        linkList.append(line.rstrip())
    f.close()
    return linkList

"""
Gets all eligible disease genes in the network (not seed genes)
"""
def get_all_genes(list, seed_genes):
    all_genes=[]
    for line in list:
        genes=line.split()
        if genes[0] not in all_genes:
            if genes[0] not in seed_genes:
                all_genes.append(genes[0])
        if genes[1] not in all_genes:
            if genes[1] not in seed_genes:
                all_genes.append(genes[1])
    return all_genes

"""
Main program
"""
seedList=get_seed_genes() #get inputs
network=get_links(sys.argv[2])
geneList=get_all_genes(network, seedList)
randomList=[] #list of randomly selected genes
for i in range(int(sys.argv[3])): #loop through specified iterations
    choice=random.choice(geneList) #choose a random gene from the network
    randomList.append(choice) #add gene to list of random genes
    geneList.remove(choice) #remove gene from the network

f=open(sys.argv[4],'w') #write to outfile
for rand in randomList:
    f.write(rand)
    f.write("\n")
f.close()