annotate recursive_agg_onefile.py @ 20:797818b4e2f6 draft

planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
author scottx611x
date Mon, 21 Nov 2016 13:00:20 -0500
parents 70d6350f982d
children 202ef48959d3
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
1 from __future__ import division, print_function
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
2 from multiprocessing import Pool
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
3 import argparse
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
4 import sys
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
5
5
4543c0fdd563 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 4
diff changeset
6 import numpy as np
10
f0d4e43f95b5 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 5
diff changeset
7 import cooler
5
4543c0fdd563 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 4
diff changeset
8 from cooler.io import CoolerAggregator
4543c0fdd563 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 4
diff changeset
9 import cooler.ice
0
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
10 import h5py
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
11
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
12
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
13 FACTOR = 2
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
14 TILESIZE = 256
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
15 N_CPU = 8
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
16
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
17 def main(infile, outfile, chunksize):
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
18 c = cooler.Cooler(infile)
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
19 binsize = c.info['bin-size']
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
20 chromtable = c.chroms()[:]
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
21 chromsizes = chromtable.set_index('name')['length']
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
22 chroms = chromtable['name'].values
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
23 lengths = chromtable['length'].values
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
24 total_length = np.sum(chromsizes.values)
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
25 n_tiles = total_length / binsize / TILESIZE
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
26 n_zooms = int(np.ceil(np.log2(n_tiles)))
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
27
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
28 print(
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
29 "Copying base matrix to level {0} and producing {0} zoom levels starting from 0...".format(n_zooms),
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
30 file=sys.stderr
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
31 )
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
32
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
33 # transfer base matrix
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
34 with h5py.File(outfile, 'w') as dest, \
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
35 h5py.File(infile, 'r') as src:
20
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
36 zoomLevel = str(n_zooms)
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
37 src.copy('/', dest, zoomLevel)
0
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
38 print(zoomLevel, file=sys.stderr)
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
39
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
40
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
41 # produce aggregations
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
42 with h5py.File(outfile, 'r+') as f:
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
43 grp = f[str(n_zooms)]
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
44 c = cooler.Cooler(grp)
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
45 binsize = cooler.info(grp)['bin-size']
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
46
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
47 for i in range(n_zooms - 1, -1, -1):
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
48 zoomLevel = str(i)
19
70d6350f982d planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 18
diff changeset
49
20
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
50 # aggregate
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
51 new_binsize = binsize * FACTOR
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
52 new_bins = cooler.util.binnify(chromsizes, new_binsize)
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
53
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
54 reader = CoolerAggregator(c, new_bins, chunksize)
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
55
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
56 grp = f.create_group(zoomLevel)
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
57 f.attrs[zoomLevel] = new_binsize
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
58 cooler.io.create(grp, chroms, lengths, new_bins, reader)
0
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
59
20
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
60 # balance
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
61 #with Pool(N_CPU) as pool:
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
62 too_close = 20000 # for HindIII
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
63 ignore_diags = max(int(np.ceil(too_close / new_binsize)), 3)
0
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
64
20
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
65 bias = cooler.ice.iterative_correction(
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
66 f, zoomLevel,
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
67 chunksize=chunksize,
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
68 min_nnz=10,
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
69 mad_max=3,
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
70 ignore_diags=ignore_diags,
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
71 map=map)
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
72 h5opts = dict(compression='gzip', compression_opts=6)
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
73 grp['bins'].create_dataset('weight', data=bias, **h5opts)
0
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
74
20
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
75 print(zoomLevel, file=sys.stderr)
0
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
76
20
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
77 c = cooler.Cooler(grp)
797818b4e2f6 planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents: 19
diff changeset
78 binsize = new_binsize
0
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
79
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
80 if __name__ == '__main__':
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
81 parser = argparse.ArgumentParser(
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
82 description="Recursively aggregate a single resolution cooler file into a multi-resolution file.")
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
83 parser.add_argument(
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
84 "cooler_file",
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
85 help="Cooler file",
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
86 metavar="COOLER_PATH")
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
87 parser.add_argument(
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
88 "--out", "-o",
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
89 help="Output text file")
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
90 args = vars(parser.parse_args())
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
91
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
92
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
93 infile = args['cooler_file']
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
94 if args['out'] is None:
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
95 outfile = infile.replace('.cool', '.multires.cool')
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
96 else:
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
97 outfile = args['out']
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
98
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
99 chunksize = int(1e6)
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
100 main(infile, outfile, chunksize)
c7d624661e8a planemo upload commit 9e980862c8b763cf3dd0209be49f0d2d75a90614-dirty
scottx611x
parents:
diff changeset
101