def sample_sampling_diversity(sample, samdir, args=None):
# Sampling sample to "size, and compute the "index" of the subsample
assert len(args) == 3
size = args[0]
indices = args[1]
workdir = args[2]
if size is None: # no sampling, compute diversity of original sample
subsampledir = samdir
else:
subsampledir = os.path.join(workdir, "subsample")
system("mkdir -p %s" % subsampledir)
libsample.sampling(sample, samdir, subsampledir, (size,))
sampling = SampleDiversityStat()
sampling.set_sample_info(sample)
counts = sample_get_clone_attrs(sample.name, subsampledir, "count")
#import rpy2.rinterface as rinterface
#rinterface.set_initoptions(('rpy2', '--no-save', '--no-restore'))
#rinterface.initr()
import rpy2.robjects as robjs
from rpy2.robjects.packages import importr
vegan = importr("vegan")
rcounts = robjs.IntVector(counts)
for index in indices:
if index == 'numclone':
sampling[index] = len(counts)
elif index == 'fisher_alpha':
rfisher = vegan.fisher_alpha(rcounts)
sampling[index] = rfisher[0]
else:
rval = vegan.diversity(rcounts, index)
sampling[index] = rval[0]
return sampling
def test_sampling(self):
self.assertRaises(ValueError, libsample.sampling, None, [5])
empty_sample = libsample.Sample("Empty", [])
self.assertRaises(ValueError, libsample.sampling, empty_sample, [7])
self.assertRaises(ValueError, libsample.sampling, self.sample, [0])
self.assertRaises(ValueError, libsample.sampling, self.sample, [20])
subsample = libsample.sampling(self.sample, [7])
self.assertTrue(len(subsample.clones) <= 4)
self.assertEqual(subsample.size, 7)
size = sum([c.count for c in subsample.clones])
self.assertEqual(size, 7)
freq = sum([c.freq for c in subsample.clones])
self.assertTrue(abs(1 - freq) < 0.01)
seqs = [c.nuc for c in subsample.clones]
uniqseqs = []
for s in seqs:
if s not in uniqseqs:
uniqseqs.append(s)
self.assertTrue(len(uniqseqs) == len(seqs))
for c in subsample.clones:
self.assertTrue(c.count > 0)
self.assertTrue(c.freq > 0)
self.assertTrue(c.nuc in self.s2v)
self.assertTrue(c.vgenes == self.s2v[c.nuc])
self.assertTrue(c.jgenes == self.s2j[c.nuc])
