decontaminator: utils/preprocess.py annotate

annotate utils/preprocess.py @ 0:b856d3d95413 draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72

author	iuc
date	Mon, 09 Jan 2023 13:27:09 +0000
parents
children

rev	line source
0 b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	1 #!/usr/bin/env python
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	2 # -- coding: utf-8 --
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	3 # Credits: Grigorii Sukhorukov, Macha Nikolski
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	4
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	5 import math
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	6 import os
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	7 import pathlib
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	8 import random
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	9
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	10 import h5py
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	11 import numpy as np
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	12 from Bio import SeqIO
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	13 from Bio.Seq import Seq
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	14 from Bio.SeqRecord import SeqRecord
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	15 from sklearn.utils import shuffle
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	16
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	17
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	18 def reverse_complement(fragment):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	19 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	20 provides reverse complement to sequences
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	21 Input:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	22 sequences - list with SeqRecord sequences in fasta format
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	23 Output:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	24 complementary_sequences -
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	25 list with SeqRecord complementary sequences in fasta format
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	26 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	27 # complementary_sequences = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	28 # for sequence in sequences:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	29 # complementary_sequence = SeqRecord(
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	30 # seq=Seq(sequence.seq).reverse_complement(),
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	31 # id=sequence.id + "_reverse_complement",
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	32 # )
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	33 # complementary_sequences.append(complementary_sequence)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	34 fragment = fragment[::-1].translate(str.maketrans('ACGT', 'TGCA'))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	35 return fragment
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	36
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	37
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	38 def introduce_mutations(seqs, mut_rate, rs=None):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	39 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	40 Function that mutates sequences in the entering fasta file
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	41 A proportion of nucleotides are changed to other nucleotide
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	42 Not yet taking account of mutation for gaps
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	43 mut_rate - proportion from 0.0 to 1.0, float
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	44 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	45 random.seed(a=rs)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	46 assert 0.0 <= mut_rate <= 1.0
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	47 mutated_seqs = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	48 for seq in seqs:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	49 mut_seq = list(str(seq.seq))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	50 l_ = len(mut_seq)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	51 mutated_sites_i = random.sample(range(l_), int(mut_rate * l_))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	52 for mut_site_i in mutated_sites_i:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	53 mut_site = mut_seq[mut_site_i]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	54 mutations = ["A", "C", "T", "G"]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	55 if mut_site in mutations:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	56 mutations.remove(mut_site)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	57 mut_seq[mut_site_i] = random.sample(mutations, 1)[0]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	58 mutated_seq = SeqRecord(
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	59 seq=Seq("".join(mut_seq)),
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	60 id=seq.id + f"mut_{mut_rate}",
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	61 name="",
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	62 description="",
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	63 )
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	64 mutated_seqs.append(mutated_seq)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	65 return mutated_seqs
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	66
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	67
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	68 def separate_by_length(length_, seq_list, fold=None,):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	69 # TODO: add docs
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	70 included = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	71 to_process = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	72 excluded = 0
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	73 for seq_ in seq_list:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	74 l_ = len(seq_.seq)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	75 if l_ >= length_:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	76 if fold is None:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	77 included.append(seq_)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	78 elif l_ < length_ * fold:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	79 included.append(seq_)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	80 else:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	81 to_process.append(seq_)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	82 else:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	83 excluded += 1
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	84 print(f"A total of {excluded} sequences was excluded due to being smaller than {length_}")
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	85 return included, to_process
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	86
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	87
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	88 def chunks(lst, n):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	89 """Yield successive n-sized chunks from lst.
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	90 https://stackoverflow.com/questions/312443/how-do-you-split-a-list-into-evenly-sized-chunks"""
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	91 for i in range(0, len(lst), n):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	92 yield lst[i:i + n]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	93
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	94
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	95 def correct(frag):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	96 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	97 leaves only unambiguous DNA code (ACTG-)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	98 Input:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	99 frag - string of nucleotides
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	100 Output:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	101 pr_frag - corrected string of nucleotides
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	102 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	103 pr_frag = frag.upper()
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	104 pr_frag_s = set(pr_frag)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	105 if pr_frag_s != {"A", "C", "G", "T", "-"}:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	106 for letter in pr_frag_s - {"A", "C", "G", "T", "-"}:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	107 pr_frag = pr_frag.replace(letter, "-")
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	108 return pr_frag
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	109
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	110
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	111 def fragmenting(sequences, sl_wind_size, max_gap=0.05, sl_wind_step=None):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	112 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	113 slices sequences in fragments by sliding window
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	114 based on its size and step.
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	115 last fragment is padded by '-'
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	116 fragments have ambiguous bases replaced by '-'
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	117 fragments with many '-' are discarded
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	118 Input:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	119 sequences - list with SeqRecord sequences in fasta format
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	120 max_gap - max allowed proportion of '-'
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	121 sl_wind_size - sliding window step
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	122 sl_wind_step - sliding window step, by default equals
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	123 sliding window size (None is replaced by it)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	124 Output:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	125 fragments - list with sequence fragments
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	126 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	127 if sl_wind_step is None:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	128 sl_wind_step = sl_wind_size
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	129 fragments = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	130 fragments_rc = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	131 out_sequences = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	132 for sequence in sequences:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	133 seq = str(sequence.seq)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	134 n_fragments = 1 + max(0, math.ceil((len(seq) - sl_wind_size) / sl_wind_step))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	135 for n in range(n_fragments):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	136 if n + 1 != n_fragments:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	137 frag = seq[n * sl_wind_step: n * sl_wind_step + sl_wind_size]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	138 elif n_fragments == 1:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	139 # padding the shorter fragment to sl_wind_size
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	140 frag_short = seq[n * sl_wind_step: n * sl_wind_step + sl_wind_size]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	141 frag = frag_short + (sl_wind_size - len(frag_short)) * "-"
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	142 else:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	143 frag = seq[(len(seq) - sl_wind_size):]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	144 # replace ambiguous characters
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	145 frag = correct(frag)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	146 assert len(frag) == sl_wind_size, f"{len(frag)} vs {sl_wind_size}"
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	147 # skipping sequences with many gaps
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	148 if frag.count("-") / sl_wind_size <= max_gap:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	149 fragments.append(frag)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	150 # generating reverse complement
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	151 fragments_rc.append(reverse_complement(frag))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	152 fr_seq = SeqRecord(
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	153 seq=Seq(frag),
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	154 id=f"{sequence.id}_{n*sl_wind_step}_{sl_wind_size}",
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	155 name="",
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	156 description="",
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	157 )
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	158 out_sequences.append(fr_seq)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	159 return fragments, fragments_rc, out_sequences
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	160
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	161
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	162 def label_fasta_fragments(sequences, label):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	163 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	164 Provides labels to generated fragments stored in fasta
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	165 Input:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	166 sequences - list with SeqRecord sequences
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	167 label - type of label (bacteria, virus, plant)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	168 Output:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	169 labeled_fragments - list with labeled SeqRecord sequences
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	170 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	171 # assert label in ["virus", "plant", "bacteria"]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	172 labeled_fragments = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	173 for sequence in sequences:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	174 sequence.id = sequence.id + f"_{label}"
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	175 labeled_fragments.append(sequence)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	176 return labeled_fragments
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	177
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	178
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	179 def one_hot_encode(fragments):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	180 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	181 produces one-hot matrices from fragments and labels
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	182 '-' is given all zeros
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	183 Input:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	184 fragments - list with sequence fragments
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	185 label - type of label (int <= depth)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	186 label_depth - number of possible labels
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	187 Output:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	188 encoded_fragments - list with one-hot encoded fragments
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	189 labels - list with one-hot encoded labels
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	190 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	191 import tensorflow as tf
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	192 encoded_fragments = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	193 map_dict = {"A": 0, "C": 1, "G": 2, "T": 3, "-": -1}
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	194 for frag in fragments:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	195 frag_array = np.array(list(frag))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	196 integer_encoded = np.int8(np.vectorize(map_dict.get)(frag_array))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	197 one_hot_encoded = tf.one_hot(integer_encoded, depth=4, dtype=tf.int8).numpy()
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	198 encoded_fragments.append(one_hot_encoded)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	199 encoded_fragments = np.stack(encoded_fragments)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	200 return encoded_fragments
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	201
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	202
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	203 def prepare_labels(fragments, label, label_depth):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	204 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	205 produces one-hot labels
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	206 '-' is given all zeros
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	207 Input:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	208 fragments - list with sequence fragments
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	209 label - type of label (int <= depth)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	210 label_depth - number of possible labels
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	211 Output:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	212 labels - list with one-hot encoded labels
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	213 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	214 import tensorflow as tf
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	215 n_fragments = len(fragments)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	216 labels = np.int8(np.full(n_fragments, label))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	217 labels = tf.one_hot(labels, depth=label_depth).numpy()
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	218 return labels
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	219
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	220
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	221 # TODO: write docs for functions
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	222 def calculate_total_length(seq_path):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	223 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	224 Calculate total length of the sequences in the fasta file.
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	225 Needed for weighted sampling
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	226 Input:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	227 seq_path - path to the file with sequences
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	228 Output:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	229 seq_length - total length of all sequences in the file
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	230 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	231 seqs = list(SeqIO.parse(seq_path, "fasta"))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	232 seq_length = 0
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	233 for seq in seqs:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	234 seq_length += len(seq.seq)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	235 return seq_length
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	236
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	237
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	238 def prepare_seq_lists(in_paths, n_fragments, weights=None,):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	239 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	240 selects files with sequences based on extension
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	241 and calculates number of fragments to be sampled
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	242 Input:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	243 in_paths - list of paths to folder with sequence files. Can be a string also a string
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	244 n_fragments - number of fragments to be sampled
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	245 weights - upsampling of fragments. fractions should sum to one
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	246 Output:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	247 seqs_list - list with path to files with sequences
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	248 n_fragments_list - number of fragments to be sampled
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	249 lists are zipped to work with ray iterators
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	250 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	251 # case when we recieve a single sequence file
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	252 if type(in_paths) is str and in_paths.endswith(('.fna', '.fasta')):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	253 return [[in_paths, n_fragments]]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	254 else:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	255 # transform string to list
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	256 if type(in_paths) is str or type(in_paths) is pathlib.PosixPath:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	257 in_paths = [in_paths]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	258
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	259 if weights:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	260 assert len(weights) == len(in_paths)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	261 assert 1.01 > round(sum(weights), 2) > 0.99
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	262 else:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	263 l_ = len(in_paths)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	264 weights = [1 / l_] * l_
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	265 n_fragments_list_all = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	266 seqs_list_all = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	267 for in_paths, w_ in zip(in_paths, weights):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	268 seqs_list = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	269 seq_length_list = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	270 total_length = 0
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	271 for file in os.listdir(in_paths):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	272 if file.endswith("fna") or file.endswith("fasta"):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	273 seq_path = (os.path.join(in_paths, file))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	274 seqs_length = calculate_total_length(seq_path)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	275 seqs_list.append(seq_path)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	276 seq_length_list.append(seqs_length)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	277 total_length += seqs_length
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	278 # + 1 may lead to a slightly bigger number than desired
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	279 n_fragments_list = [((seq_length / total_length) * n_fragments * w_ + 1) for seq_length in seq_length_list]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	280 n_fragments_list_all.extend(n_fragments_list)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	281 seqs_list_all.extend(seqs_list)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	282 print("list calculation done")
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	283 return list(zip(seqs_list_all, n_fragments_list_all))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	284
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	285
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	286 def sample_fragments(seq_container, length, random_seed=1, limit=None, max_gap=0.05, sl_wind_step=None):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	287 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	288 Randomly samples fragments from sequences in the list.
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	289 Is a bit cumbersome written to work with ray.
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	290 Input:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	291 seq_container - list with each entry containing path to sequence,
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	292 and n samples from this sequence.
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	293 length - desired length of sampled fragments
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	294 Output:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	295 fragments - list with sequence fragments
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	296 """
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	297 random.seed(a=random_seed)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	298 total_fragments = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	299 total_fragments_rc = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	300 total_seqs = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	301 for entry in seq_container:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	302 seq = list(SeqIO.parse(entry[0], "fasta"))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	303 n_fragments = entry[1]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	304 seqs = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	305 fragments = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	306 fragments_rc = []
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	307 counter_1 = 0
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	308 counter_2 = 0
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	309 while counter_1 < n_fragments:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	310 # select chromosomes if there are any
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	311 fragment_full = random.choice(seq)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	312 r_end = len(fragment_full.seq) - length
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	313 try:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	314 r_start = random.randrange(r_end)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	315 fragment = SeqRecord(
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	316 seq=fragment_full.seq[r_start:(r_start + length)],
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	317 id=f"{fragment_full.id}_{length}_{r_start}",
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	318 name="",
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	319 description="",
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	320 )
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	321 temp_, temp_rc, _ = fragmenting([fragment], length, max_gap, sl_wind_step=sl_wind_step)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	322 if temp_ and temp_rc:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	323 seqs.append(fragment)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	324 fragments.extend(temp_)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	325 fragments_rc.extend(temp_rc)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	326 counter_1 += 1
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	327 except ValueError:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	328 # print(f"{fragment_full.id} has length {len(fragment_full.seq)} and is too short to be sampled")
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	329 pass
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	330 counter_2 += 1
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	331 if limit:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	332 assert counter_2 <= limit * n_fragments, f"While cycle iterated more than {limit}, data is ambiguous." \
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	333 f" Only {len(fragments)} fragments were sampled out of {n_fragments}"
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	334 total_fragments.extend(fragments)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	335 total_fragments_rc.extend(fragments_rc)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	336 total_seqs.extend(seqs)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	337 # print("sequence sampling done")
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	338 return total_fragments, total_fragments_rc, total_seqs
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	339
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	340
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	341 def prepare_ds_fragmenting(in_seq, label, label_int, fragment_length, sl_wind_step, max_gap=0.05, n_cpus=1):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	342 if sl_wind_step is None:
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	343 sl_wind_step = int(fragment_length / 2)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	344 # generating viral fragments and labels
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	345 seqs = list(SeqIO.parse(in_seq, "fasta"))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	346 frags, frags_rc, seqs_ = fragmenting(seqs, fragment_length, max_gap=max_gap, sl_wind_step=sl_wind_step)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	347 encoded = one_hot_encode(frags)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	348 encoded_rc = one_hot_encode(frags_rc)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	349 labs = prepare_labels(frags, label=label_int, label_depth=2)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	350 seqs_ = label_fasta_fragments(seqs_, label=label)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	351 # subsetting to unique fragments
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	352 u_encoded, indices = np.unique(encoded, axis=0, return_index=True)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	353 u_encoded_rc = encoded_rc[indices]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	354 u_labs = labs[indices]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	355 u_seqs = [seqs_[i] for i in indices]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	356 assert (np.shape(u_encoded)[0] == np.shape(u_encoded_rc)[0])
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	357 print(f"Encoding {label} sequences finished")
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	358 # print(f"{np.shape(u_encoded)[0]} forward fragments generated")
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	359 n_frags = np.shape(u_encoded)[0]
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	360 return u_encoded, u_encoded_rc, u_labs, u_seqs, n_frags
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	361
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	362
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	363 def prepare_ds_sampling(in_seqs, fragment_length, n_frags, label, label_int, random_seed, n_cpus=1, limit=100):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	364 # generating plant fragments and labels
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	365 seqs_list = prepare_seq_lists(in_seqs, n_frags)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	366 frags, frags_rc, seqs_ = sample_fragments(seqs_list, fragment_length, random_seed, limit=limit, max_gap=0.05)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	367 frags, frags_rc, seqs_ = shuffle(frags, frags_rc, seqs_, random_state=random_seed, n_samples=int(n_frags))
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	368 encoded = one_hot_encode(frags)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	369 encoded_rc = one_hot_encode(frags_rc)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	370 labs = prepare_labels(frags, label=label_int, label_depth=2)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	371 seqs_ = label_fasta_fragments(seqs_, label=label)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	372 assert (np.shape(encoded)[0] == np.shape(encoded_rc)[0])
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	373 print(f"Encoding {label} sequences finished")
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	374 # print(f"{np.shape(encoded)[0]} forward fragments generated")
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	375 return encoded, encoded_rc, labs, seqs_, n_frags
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	376
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	377
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	378 def storing_encoded(encoded, encoded_rc, labs, out_path, ):
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	379 f = h5py.File(out_path, "w")
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	380 f.create_dataset("fragments", data=encoded)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	381 f.create_dataset("fragments_rc", data=encoded_rc)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	382 f.create_dataset("labels", data=labs)
b856d3d95413 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/decontaminator commit 3f8e87001f3dfe7d005d0765aeaa930225c93b72 iuc parents: diff changeset	383 f.close()

Mercurial > repos > iuc > decontaminator

annotate utils/preprocess.py @ 0:b856d3d95413 draft default tip