#!/usr/bin/env python
# coding: utf-8

# In[33]:


from __future__ import print_function, division


# In[1]:


get_ipython().system('pwd')


# In[2]:


get_ipython().system('ls')


# In[4]:


get_ipython().system('cat README.txt')
get_ipython().system('cat TFs_list.txt')


# In[34]:


import glob

positives_loc = "/mnt/data/ENCODE/peaks_spp/mar2012/distinct/idrOptimalBlackListFilt"
ambig_loc = "/mnt/data/ENCODE/peaks_spp/mar2012/distinct/pooledPseudoReps/regionPeak"

tfs_to_pursue = []

for a_tf in open("TFs_list.txt"):
    a_tf = a_tf.rstrip()
    a_tf = a_tf[0]+a_tf[1:].lower()
    
    to_glob = positives_loc+"/*"+a_tf+"*"
    
    matching_files_pos = glob.glob(positives_loc+"/*"+a_tf+"*")
     
    if (len(matching_files_pos) > 1):
        tfs_to_pursue.append((a_tf,matching_files_pos,glob.glob(ambig_loc+"/*"+a_tf+"*")))
        print(a_tf)
        print(len(matching_files_pos))
    else:
        print("Skipping:",a_tf, len(matching_files_pos))
    
    #print(glob.glob(positives_loc+"/*"+a_tf+"*"))
    #to_glob = ambig_loc+"/*"+a_tf+"*"
    #print(glob.glob(ambig_loc+"/*"+a_tf+"*"))
    #print(len(glob.glob(ambig_loc+"/*"+a_tf+"*")))


# In[31]:


tfs_to_pursue


# In[ ]: