python/paths_load_data_pickle.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3
   4 import cPickle
   5
   6 def paths_load_data_pickle(expt,genome_info,PAR):
   7    """
   8
   9    """
  10
  11    # function [Sequences, Acceptors, Donors, Exons, Ests, Noises] = paths_load_data(expt,genome_info,PAR)
  12    # Load the relevant file and return the alignment data
  13
  14    # expt can be 'training','validation' or 'test'
  15
  16    assert expt in ['training','validation','test']
  17
  18    tmp_dir = '/fml/ag-raetsch/home/fabio/tmp'
  19
  20    Noises = [];
  21
  22    if expt == 'training':
  23       if PAR.microexon:
  24          if PAR.LOCAL_ALIGN: # local version
  25
  26             train_data = '%s/microexon_train_data.pickle' % genome_info.basedir
  27             data = cPickle.load(open(train_data))
  28
  29          else: # global version
  30             pass
  31
  32
  33       else:
  34          train_data = '%s/exons_train_local.pickle' % genome_info.basedir
  35          data = cPickle.load(open(train_data))
  36
  37       print 'train_data is %s' % train_data
  38
  39       Sequences = data['Train']       # dna sequences
  40       Acceptors = data['TrainAcc']    # acceptor scores
  41       Donors    = data['TrainDon']    # donor scores
  42       Exons     = data['TrainExon']   # exon boundaries
  43       Ests      = data['TrainEsts']   # est sequences
  44
  45    # Lower all indices by one to convert matlab
  46    # to python indices
  47
  48    Exons -= 1
  49
  50    return Sequences, Acceptors, Donors, Exons, Ests, Noises