"""Script to create family scores dictionaries """ import sys import cPickle from gPy.Parameters import SqliteFactor from gPy.IO import read_csv pa_size_lim_value = 3 precision_value = 1 for f in sys.argv[1:]: if f.endswith('.data'): stem = f[:-5] fobj = open(f) elif f.endswith('.data.gz'): stem = f[:-8] import gzip fobj = gzip.open(f) else: raise ValueError("Input files must end with either '.data' or '.data.gz'") data = SqliteFactor.init_rawdata(read_csv(fobj),dbfilename=stem+'.db') cPickle.dump(data,open(stem+'.sqldata.pck','w')) family_scores = data.make_family_scores_naively(pa_size_lim=pa_size_lim_value,precision=precision_value) cPickle.dump(family_scores,open('%s_family_scores_%s_%s.pck' % (stem,pa_size_lim_value,precision_value),'w'))