In [1]:

import cPickle as pickle
import networkx as nx

In [2]:

train_cont_sc_1_3_graphs = []
train_cont_sc_1_3_labels = []
# labels:  linear = 0, lineage = 1

In [3]:

from os import listdir
from os.path import isfile, join

In [4]:

linear_continuity_files = [f for f in listdir("train-linear/continuity") if isfile(join("train-linear/continuity", f))]
lineage_continuity_files = [f for f in listdir("train-lineage/continuity") if isfile(join("train-lineage/continuity", f))]

In [5]:

for f in linear_continuity_files:
    path = "train-linear/continuity/" + f
    try:
        g = nx.read_gml(path)
    except:
        print "bad file format: %s" % f
    train_cont_sc_1_3_graphs.append(g)
    train_cont_sc_1_3_labels.append(0)
    
for f in lineage_continuity_files:
    path = "train-lineage/continuity/" + f
    try:
        g = nx.read_gml(path)
    except:
        print "bad file format: %s" % f
    train_cont_sc_1_3_graphs.append(g)
    train_cont_sc_1_3_labels.append(1)

  edge [
  ^
Expected "]" (at char 6318), (line:351, col:3)
bad file format: 5981e54a-d77c-11e5-b052-086266a2412a-0-sampled-500-slicestratified-0.12-resample-0-0.1-unimodal-filtered-minmax-by-weight-continuity.gml

In [6]:

pickle.dump(train_cont_sc_1_3_graphs,open("train-cont-sc-1-3-graphs.pkl",'wb'))
pickle.dump(train_cont_sc_1_3_labels,open("train-cont-sc-1-3-labels.pkl",'wb'))

In [ ]: