Mercurial > repos > jay > pdaug_tsvtofasta

annotate PDAUG_Peptide_Ngrams/PDAUG_Peptide_Ngrams.py @ 4:87b77f2ddb0c draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 8b18552f6d2b2261efebe1075ff4c18a295b94dd"
author jay
date Tue, 29 Dec 2020 18:13:00 +0000
parents c3f0b3a6339e
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
1 import matplotlib
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
2 matplotlib.use('Agg')
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
3 import os
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
4 import sys
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
5 sys.path.insert(0, os.path.abspath('..'))
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
6 import quantiprot
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
7 from quantiprot.utils.io import load_fasta_file
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
8 from quantiprot.utils.feature import Feature, FeatureSet
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
9 from quantiprot.metrics.aaindex import get_aa2hydropathy
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
10 from quantiprot.metrics.basic import identity
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
11 from quantiprot.metrics.ngram import pattern_match, pattern_count
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
12 from quantiprot.analysis.ngram import ngram_count
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
13 from quantiprot.analysis.ngram import zipf_law_fit
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
14 from matplotlib import pyplot as plt
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
15
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
16
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
17 def Run_ngrams(fasta1, fasta2, OutFile ):
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
18
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
19 alphasyn_seq = load_fasta_file(fasta1)
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
20 amyload_pos_seq = load_fasta_file(fasta2)
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
21
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
22 fs_aa = FeatureSet("aa patterns")
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
23 fs_aa.add(identity)
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
24 fs_aa.add(pattern_match, pattern='VT', padded=True)
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
25 fs_aa.add(pattern_count, pattern='VT')
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
26
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
27 result_seq = fs_aa(alphasyn_seq)
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
28
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
29 fs_hp = FeatureSet("hydropathy patterns")
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
30 fs_hp.add(Feature(get_aa2hydropathy()))
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
31 fs_hp.add(Feature(get_aa2hydropathy()).then(pattern_match, pattern=[0.0, 2.0],
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
32 metric='taxi', radius=1.0))
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
33 result_seq2 = fs_hp(alphasyn_seq)
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
34 result_freq = ngram_count(alphasyn_seq, n=2)
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
35 result_fit = zipf_law_fit(amyload_pos_seq, n=3, verbose=True)
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
36
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
37 counts = sorted(result_fit["ngram_counts"], reverse=True)
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
38 ranks = range(1, len(counts)+1)
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
39
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
40 slope = result_fit["slope"]
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
41 harmonic_num = sum([rank**-slope for rank in ranks])
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
42 fitted_counts = [(rank**-slope) / harmonic_num * sum(counts) for rank in ranks]
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
43
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
44 plt.plot(ranks, counts, 'k', label="empirical")
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
45 plt.plot(ranks, fitted_counts, 'k--',
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
46 label="Zipf's law\nslope: {:.2f}".format((slope)))
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
47 plt.xlabel('rank')
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
48 plt.ylabel('count')
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
49 plt.xscale('log')
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
50 plt.yscale('log')
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
51 plt.legend()
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
52
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
53 plt.savefig(OutFile)
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
54
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
55 if __name__=="__main__":
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
56
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
57
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
58 import argparse
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
59
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
60 parser = argparse.ArgumentParser()
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
61
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
62 parser.add_argument("-f1", "--Fasta1",
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
63 required=True,
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
64 default=None,
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
65 help="First fasta file")
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
66
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
67 parser.add_argument("-f2", "--Fasta2",
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
68 required=True,
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
69 default=None,
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
70 help="Second fasta file")
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
71
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
72
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
73 parser.add_argument("--OutFile",
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
74 required=True,
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
75 help="HTML out file",
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
76 default="report.html")
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
77
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
78
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
79 args = parser.parse_args()
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
80
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
81 Run_ngrams(args.Fasta1, args.Fasta2, args.OutFile)
c3f0b3a6339e "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
82