def make_pass(stressor):
#print "Parsing data for '%s'...\n" % stressor
experiments = [ DataSet.load(input % stressor) for input in inputs ]
reference = experiments[0]
for data in experiments:
# Find the ratio between the amount of red and green fluorescence that was
# detected. This ratio is assumed to be one for most data analysis
# purposes, so the raw data needs to be corrected.
green, red = 0, 0
for feature in data:
red += feature.signal.red.intensity
green += feature.signal.green.intensity
data.intensity_ratio = red / green
data.log_ratio = math.log(red / green, 2)
def correction(feature):
feature.log_ratio -= data.log_ratio
return feature
data.apply(correction)
for data in experiments:
for feature, zero in zip(data, reference):
feature.normed_ratio = feature.log_ratio - zero.log_ratio
for data in experiments:
def irrational(feature):
return math.isnan(feature.normed_ratio)
def noisy(feature):
return (feature.signal.red.signal_to_noise < 1) or \
(feature.signal.green.signal_to_noise < 1)
def unnamed(feature):
return feature.name in ('None', 'EMPTY')
# This filter was proposed by team JKRW.
def inconsistent(feature):
return feature.regression_quality < 0.5
data.prune(irrational)
data.prune(noisy)
data.prune(unnamed)
data.prune(inconsistent)
for data, output in zip(experiments, outputs):
print "Saving %d features for '%s'." % (len(data), stressor)
data.save(output % stressor)
print
#!/usr/bin/env python
from __future__ import division
from microarray import DataSet
inputs = [
'data/A+D.000.gpr',
'data/A+D.030.gpr',
'data/A+D.060.gpr',
'data/A+D.180.gpr' ]
for input in inputs:
data = DataSet.load(input)
print len(data)
