from __future__ import division
from timeit import timeit
from biom import load_table
import pandas as pd
from glob import glob

t = load_table('table_mc5038.biom')

t.shape

axes = ['observation', 'sample']
shape = t.shape

for i, axis in enumerate(axes):
    data[axis] = []
    step_size = np.ceil(shape[i] / 10)
    for n in arange(step_size, shape[i], step_size):
        ids = t.ids(axis=axis)[:n]
        subt = t.filter(ids, axis=axis, inplace=False)
        out_path = 'test-tables/t.%d.%s.biom' % (n, axis)
        subt.to_json('me', open(out_path, 'w'))

data = {'observation': [], 'sample': []}
for fp in glob('test-tables/t.*.biom'):
    fields = fp.split('.')
    n = fields[1]
    axis = fields[2]
    def sr():
            !single_rarefaction.py -i $fp -o x.biom -d 500
    data[axis].append((int(n), timeit(sr, number=3) / 3))

df = pd.DataFrame(data['observation'], columns=['n', 'mean run time'], dtype=float)
df.sort('n', inplace=True)
df.plot(x='n', y='mean run time')

df = pd.DataFrame(data['sample'], columns=['n', 'mean run time'], dtype=float)
df.sort('n', inplace=True)
df.plot(x='n', y='mean run time', legend=False)

df.plot?