Mercurial > repos > iuc > virhunter

annotate utils/preprocess.py @ 2:ea2cccb9f73e draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit c3685ed6a70b47012b62b95a2a3db062bd3b7475
author iuc
date Thu, 05 Jan 2023 14:27:54 +0000
parents 457fd8fd681a
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
1 #!/usr/bin/env python
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
2 # -*- coding: utf-8 -*-
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
3 # Credits: Grigorii Sukhorukov, Macha Nikolski
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
4 import math
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
5 import os
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
6 import pathlib
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
7 import random
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
8
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
9 import h5py
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
10 import numpy as np
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
11 from Bio import SeqIO
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
12 from Bio.Seq import Seq
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
13 from Bio.SeqRecord import SeqRecord
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
14 from sklearn.utils import shuffle
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
15
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
16
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
17 def reverse_complement(fragment):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
18 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
19 provides reverse complement to sequences
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
20 Input:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
21 sequences - list with SeqRecord sequences in fasta format
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
22 Output:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
23 complementary_sequences -
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
24 list with SeqRecord complementary sequences in fasta format
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
25 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
26 # complementary_sequences = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
27 # for sequence in sequences:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
28 # complementary_sequence = SeqRecord(
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
29 # seq=Seq(sequence.seq).reverse_complement(),
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
30 # id=sequence.id + "_reverse_complement",
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
31 # )
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
32 # complementary_sequences.append(complementary_sequence)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
33 fragment = fragment[::-1].translate(str.maketrans('ACGT', 'TGCA'))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
34 return fragment
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
35
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
36
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
37 def introduce_mutations(seqs, mut_rate, rs=None):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
38 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
39 Function that mutates sequences in the entering fasta file
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
40 A proportion of nucleotides are changed to other nucleotide
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
41 Not yet taking account of mutation for gaps
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
42 mut_rate - proportion from 0.0 to 1.0, float
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
43 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
44 random.seed(a=rs)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
45 assert 0.0 <= mut_rate <= 1.0
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
46 mutated_seqs = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
47 for seq in seqs:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
48 mut_seq = list(str(seq.seq))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
49 l_ = len(mut_seq)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
50 mutated_sites_i = random.sample(range(l_), int(mut_rate * l_))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
51 for mut_site_i in mutated_sites_i:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
52 mut_site = mut_seq[mut_site_i]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
53 mutations = ["A", "C", "T", "G"]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
54 if mut_site in mutations:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
55 mutations.remove(mut_site)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
56 mut_seq[mut_site_i] = random.sample(mutations, 1)[0]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
57 mutated_seq = SeqRecord(
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
58 seq=Seq("".join(mut_seq)),
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
59 id=seq.id + f"mut_{mut_rate}",
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
60 name="",
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
61 description="",
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
62 )
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
63 mutated_seqs.append(mutated_seq)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
64 return mutated_seqs
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
65
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
66
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
67 def separate_by_length(length_, seq_list, fold=None,):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
68 # TODO: add docs
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
69 included = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
70 to_process = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
71 excluded = 0
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
72 for seq_ in seq_list:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
73 l_ = len(seq_.seq)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
74 if l_ >= length_:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
75 if fold is None:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
76 included.append(seq_)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
77 elif l_ < length_ * fold:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
78 included.append(seq_)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
79 else:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
80 to_process.append(seq_)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
81 else:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
82 excluded += 1
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
83 print(f"A total of {excluded} sequences was excluded due to being smaller than {length_}")
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
84 return included, to_process
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
85
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
86
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
87 def chunks(lst, n):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
88 """Yield successive n-sized chunks from lst.
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
89 https://stackoverflow.com/questions/312443/how-do-you-split-a-list-into-evenly-sized-chunks"""
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
90 for i in range(0, len(lst), n):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
91 yield lst[i:i + n]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
92
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
93
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
94 def correct(frag):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
95 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
96 leaves only unambiguous DNA code (ACTG-)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
97 Input:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
98 frag - string of nucleotides
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
99 Output:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
100 pr_frag - corrected string of nucleotides
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
101 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
102 pr_frag = frag.upper()
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
103 pr_frag_s = set(pr_frag)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
104 if pr_frag_s != {"A", "C", "G", "T", "-"}:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
105 for letter in pr_frag_s - {"A", "C", "G", "T", "-"}:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
106 pr_frag = pr_frag.replace(letter, "-")
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
107 return pr_frag
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
108
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
109
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
110 def fragmenting(sequences, sl_wind_size, max_gap=0.05, sl_wind_step=None):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
111 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
112 slices sequences in fragments by sliding window
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
113 based on its size and step.
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
114 last fragment is padded by '-'
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
115 fragments have ambiguous bases replaced by '-'
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
116 fragments with many '-' are discarded
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
117 Input:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
118 sequences - list with SeqRecord sequences in fasta format
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
119 max_gap - max allowed proportion of '-'
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
120 sl_wind_size - sliding window step
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
121 sl_wind_step - sliding window step, by default equals
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
122 sliding window size (None is replaced by it)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
123 Output:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
124 fragments - list with sequence fragments
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
125 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
126 if sl_wind_step is None:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
127 sl_wind_step = sl_wind_size
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
128 fragments = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
129 fragments_rc = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
130 out_sequences = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
131 for sequence in sequences:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
132 seq = str(sequence.seq)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
133 n_fragments = 1 + max(0, math.ceil((len(seq) - sl_wind_size) / sl_wind_step))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
134 for n in range(n_fragments):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
135 if n + 1 != n_fragments:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
136 frag = seq[n * sl_wind_step: n * sl_wind_step + sl_wind_size]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
137 elif n_fragments == 1:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
138 # padding the shorter fragment to sl_wind_size
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
139 frag_short = seq[n * sl_wind_step: n * sl_wind_step + sl_wind_size]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
140 frag = frag_short + (sl_wind_size - len(frag_short)) * "-"
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
141 else:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
142 frag = seq[(len(seq) - sl_wind_size):]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
143 # replace ambiguous characters
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
144 frag = correct(frag)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
145 assert len(frag) == sl_wind_size, f"{len(frag)} vs {sl_wind_size}"
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
146 # skipping sequences with many gaps
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
147 if frag.count("-") / sl_wind_size <= max_gap:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
148 fragments.append(frag)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
149 # generating reverse complement
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
150 fragments_rc.append(reverse_complement(frag))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
151 fr_seq = SeqRecord(
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
152 seq=Seq(frag),
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
153 id=f"{sequence.id}_{n*sl_wind_step}_{sl_wind_size}",
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
154 name="",
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
155 description="",
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
156 )
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
157 out_sequences.append(fr_seq)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
158 return fragments, fragments_rc, out_sequences
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
159
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
160
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
161 def label_fasta_fragments(sequences, label):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
162 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
163 Provides labels to generated fragments stored in fasta
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
164 Input:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
165 sequences - list with SeqRecord sequences
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
166 label - type of label (bacteria, virus, plant)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
167 Output:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
168 labeled_fragments - list with labeled SeqRecord sequences
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
169 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
170 assert label in ["virus", "plant", "bacteria"]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
171 labeled_fragments = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
172 for sequence in sequences:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
173 sequence.id = sequence.id + f"_{label}"
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
174 labeled_fragments.append(sequence)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
175 return labeled_fragments
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
176
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
177
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
178 def one_hot_encode(fragments):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
179 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
180 produces one-hot matrices from fragments and labels
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
181 '-' is given all zeros
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
182 Input:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
183 fragments - list with sequence fragments
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
184 label - type of label (int <= depth)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
185 label_depth - number of possible labels
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
186 Output:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
187 encoded_fragments - list with one-hot encoded fragments
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
188 labels - list with one-hot encoded labels
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
189 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
190 import tensorflow as tf
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
191 encoded_fragments = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
192 map_dict = {"A": 0, "C": 1, "G": 2, "T": 3, "-": -1}
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
193 for frag in fragments:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
194 frag_array = np.array(list(frag))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
195 integer_encoded = np.int8(np.vectorize(map_dict.get)(frag_array))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
196 one_hot_encoded = tf.one_hot(integer_encoded, depth=4, dtype=tf.int8).numpy()
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
197 encoded_fragments.append(one_hot_encoded)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
198 encoded_fragments = np.stack(encoded_fragments)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
199 return encoded_fragments
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
200
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
201
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
202 def prepare_labels(fragments, label, label_depth):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
203 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
204 produces one-hot labels
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
205 '-' is given all zeros
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
206 Input:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
207 fragments - list with sequence fragments
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
208 label - type of label (int <= depth)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
209 label_depth - number of possible labels
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
210 Output:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
211 labels - list with one-hot encoded labels
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
212 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
213 import tensorflow as tf
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
214 n_fragments = len(fragments)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
215 labels = np.int8(np.full(n_fragments, label))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
216 labels = tf.one_hot(labels, depth=label_depth).numpy()
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
217 return labels
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
218
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
219
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
220 # TODO: write docs for functions
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
221 def calculate_total_length(seq_path):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
222 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
223 Calculate total length of the sequences in the fasta file.
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
224 Needed for weighted sampling
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
225 Input:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
226 seq_path - path to the file with sequences
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
227 Output:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
228 seq_length - total length of all sequences in the file
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
229 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
230 seqs = list(SeqIO.parse(seq_path, "fasta"))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
231 seq_length = 0
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
232 for seq in seqs:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
233 seq_length += len(seq.seq)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
234 return seq_length
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
235
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
236
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
237 def prepare_seq_lists(in_paths, n_fragments, weights=None,):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
238 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
239 selects files with sequences based on extension
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
240 and calculates number of fragments to be sampled
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
241 Input:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
242 in_paths - list of paths to folder with sequence files. Can be a string also a string
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
243 n_fragments - number of fragments to be sampled
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
244 weights - upsampling of fragments. fractions should sum to one
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
245 Output:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
246 seqs_list - list with path to files with sequences
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
247 n_fragments_list - number of fragments to be sampled
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
248 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
249 # case when we recieve a single sequence file
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
250 if type(in_paths) is str and in_paths.endswith(('.fna', '.fasta')):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
251 return [[in_paths, n_fragments]]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
252 else:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
253 # transform string to list
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
254 if type(in_paths) is str or type(in_paths) is pathlib.PosixPath:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
255 in_paths = [in_paths]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
256
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
257 if weights:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
258 assert len(weights) == len(in_paths)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
259 assert 1.01 > round(sum(weights), 2) > 0.99
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
260 else:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
261 l_ = len(in_paths)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
262 weights = [1 / l_] * l_
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
263 n_fragments_list_all = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
264 seqs_list_all = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
265 for in_paths, w_ in zip(in_paths, weights):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
266 seqs_list = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
267 seq_length_list = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
268 total_length = 0
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
269 for file in os.listdir(in_paths):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
270 if file.endswith("fna") or file.endswith("fasta"):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
271 seq_path = (os.path.join(in_paths, file))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
272 seqs_length = calculate_total_length(seq_path)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
273 seqs_list.append(seq_path)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
274 seq_length_list.append(seqs_length)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
275 total_length += seqs_length
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
276 # + 1 may lead to a slightly bigger number than desired
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
277 n_fragments_list = [((seq_length / total_length) * n_fragments * w_ + 1) for seq_length in seq_length_list]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
278 n_fragments_list_all.extend(n_fragments_list)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
279 seqs_list_all.extend(seqs_list)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
280 print("list calculation done")
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
281 return list(zip(seqs_list_all, n_fragments_list_all))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
282
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
283
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
284 def sample_fragments(seq_container, length, random_seed=1, limit=None, max_gap=0.05, sl_wind_step=None):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
285 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
286 Randomly samples fragments from sequences in the list.
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
287 Input:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
288 seq_container - list with each entry containing path to sequence,
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
289 and n samples from this sequence.
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
290 length - desired length of sampled fragments
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
291 Output:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
292 fragments - list with sequence fragments
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
293 """
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
294 random.seed(a=random_seed)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
295 total_fragments = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
296 total_fragments_rc = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
297 total_seqs = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
298 for entry in seq_container:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
299 seq = list(SeqIO.parse(entry[0], "fasta"))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
300 n_fragments = entry[1]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
301 seqs = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
302 fragments = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
303 fragments_rc = []
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
304 counter_1 = 0
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
305 counter_2 = 0
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
306 while counter_1 < n_fragments:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
307 # select chromosomes if there are any
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
308 fragment_full = random.choice(seq)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
309 r_end = len(fragment_full.seq) - length
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
310 try:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
311 r_start = random.randrange(r_end)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
312 fragment = SeqRecord(
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
313 seq=fragment_full.seq[r_start:(r_start + length)],
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
314 id=f"{fragment_full.id}_{length}_{r_start}",
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
315 name="",
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
316 description="",
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
317 )
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
318 temp_, temp_rc, _ = fragmenting([fragment], length, max_gap, sl_wind_step=sl_wind_step)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
319 if temp_ and temp_rc:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
320 seqs.append(fragment)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
321 fragments.extend(temp_)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
322 fragments_rc.extend(temp_rc)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
323 counter_1 += 1
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
324 except ValueError:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
325 # print(f"{fragment_full.id} has length {len(fragment_full.seq)} and is too short to be sampled")
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
326 pass
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
327 counter_2 += 1
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
328 if limit:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
329 assert counter_2 <= limit * n_fragments, f"While cycle iterated more than {limit}, data is ambiguous." \
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
330 f" Only {len(fragments)} fragments were sampled out of {n_fragments}"
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
331 total_fragments.extend(fragments)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
332 total_fragments_rc.extend(fragments_rc)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
333 total_seqs.extend(seqs)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
334 # print("sequence sampling done")
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
335 return total_fragments, total_fragments_rc, total_seqs
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
336
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
337
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
338 def prepare_ds_fragmenting(in_seq, label, label_int, fragment_length, sl_wind_step, max_gap=0.05, n_cpus=1):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
339 if sl_wind_step is None:
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
340 sl_wind_step = int(fragment_length / 2)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
341 # generating viral fragments and labels
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
342 seqs = list(SeqIO.parse(in_seq, "fasta"))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
343 frags, frags_rc, seqs_ = fragmenting(seqs, fragment_length, max_gap=max_gap, sl_wind_step=sl_wind_step)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
344 encoded = one_hot_encode(frags)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
345 encoded_rc = one_hot_encode(frags_rc)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
346 labs = prepare_labels(frags, label=label_int, label_depth=3)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
347 seqs_ = label_fasta_fragments(seqs_, label=label)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
348 # subsetting to unique fragments
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
349 u_encoded, indices = np.unique(encoded, axis=0, return_index=True)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
350 u_encoded_rc = encoded_rc[indices]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
351 u_labs = labs[indices]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
352 u_seqs = [seqs_[i] for i in indices]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
353 assert (np.shape(u_encoded)[0] == np.shape(u_encoded_rc)[0])
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
354 print(f"Encoding {label} sequences finished")
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
355 # print(f"{np.shape(u_encoded)[0]} forward fragments generated")
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
356 n_frags = np.shape(u_encoded)[0]
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
357 return u_encoded, u_encoded_rc, u_labs, u_seqs, n_frags
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
358
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
359
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
360 def prepare_ds_sampling(in_seqs, fragment_length, n_frags, label, label_int, random_seed, n_cpus=1, limit=100):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
361 # generating plant fragments and labels
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
362 seqs_list = prepare_seq_lists(in_seqs, n_frags)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
363 frags, frags_rc, seqs_ = sample_fragments.remote(seqs_list, fragment_length, random_seed, limit=limit, max_gap=0.05)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
364 frags, frags_rc, seqs_ = shuffle(frags, frags_rc, seqs_, random_state=random_seed, n_samples=int(n_frags))
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
365 encoded = one_hot_encode(frags)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
366 encoded_rc = one_hot_encode(frags_rc)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
367 labs = prepare_labels(frags, label=label_int, label_depth=3)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
368 seqs_ = label_fasta_fragments(seqs_, label=label)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
369 assert (np.shape(encoded)[0] == np.shape(encoded_rc)[0])
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
370 print(f"Encoding {label} sequences finished")
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
371 # print(f"{np.shape(encoded)[0]} forward fragments generated")
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
372 return encoded, encoded_rc, labs, seqs_, n_frags
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
373
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
374
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
375 def storing_encoded(encoded, encoded_rc, labs, out_path, ):
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
376 f = h5py.File(out_path, "w")
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
377 f.create_dataset("fragments", data=encoded)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
378 f.create_dataset("fragments_rc", data=encoded_rc)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
379 f.create_dataset("labels", data=labs)
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
380 f.close()
457fd8fd681a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/VirHunter commit 628688c1302dbf972e48806d2a5bafe27847bdcc
iuc
parents:
diff changeset
381 print(f"encoded fragments and labels stored in {out_path}")