analysis_version, datasetabbrev)):
continue
with open('datasets_in_progress_{0}/{1}.txt'.format(
analysis_version, datasetabbrev),
mode='wt',
encoding="utf-8",
errors="surrogateescape") as fw:
fw.write('working on {}...'.format(datasetabbrev))
print('working on {}...'.format(datasetabbrev))
# load dataset
gene_atb = datasetselection.loaddatamatrix(
datasetpath=datasetinfo['path'],
rowname='gene',
columnname='atb',
matrixname='gene_atb_associations',
skiprows=3,
skipcolumns=3,
delimiter='\t',
dtype='float64',
getmetadata=True, # need to fix False case
getmatrix=True)
# check binary
if set(np.unique(gene_atb.matrix)) != {0, 1}:
print('warning: converting matrix to binary values')
gene_atb.matrix = gene_atb.matrix != 0
gene_atb.updatedtypeattribute()
# compute feature similarity
atb_atb = gene_atb.tosimilarity(axis=1, metric=similarity_metric)
# align with clusters
commongenes = gene_atb.rowlabels[np.in1d(gene_atb.rowlabels,
gene_cluster.rowlabels)]
if custompath not in sys.path:
sys.path.append(custompath)
del custompath, custompaths
import numpy as np
import pickle
import machinelearning.datasetselection as ds
import os
# load the data
gene_atb = ds.loaddatamatrix(
'data/original_data/gene_attribute_matrix_cleaned.txt.gz',
rowname='gene',
columnname='atb',
matrixname='gene_atb_associations',
skiprows=3,
skipcolumns=3,
delimiter='\t',
dtype='float32',
getmetadata=True, # need to fix False case
getmatrix=True)
# shuffle the data
gene_atb.reorder(np.random.permutation(gene_atb.shape[0]), 0)
gene_atb.reorder(np.random.permutation(gene_atb.shape[1]), 1)
# standardize the data
row_mean = gene_atb.matrix.mean(1)
row_stdv = gene_atb.matrix.std(1)
standardized_row_mean = (row_mean - row_mean.mean()) / row_mean.std()
standardized_row_stdv = (row_stdv - row_stdv.mean()) / row_stdv.std()