Mercurial > repos > fubar > egapx_runner

annotate nf/subworkflows/ncbi/target_proteins/miniprot/main.nf @ 0:d9c5c5b87fec draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
author fubar
date Sat, 03 Aug 2024 11:16:53 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
1 #!/usr/bin/env nextflow
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
2
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
3 nextflow.enable.dsl=2
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
4
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
5 include { merge_params } from '../../utilities'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
6
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
7
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
8 def get_effective_params(parameters, max_intron) {
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
9 def default_params = "-t 8 -G ${max_intron}"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
10 def value = parameters.get("miniprot", "")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
11 value = value.replaceFirst("-cpu-count", "-t")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
12 value = value.replaceFirst("-max-intron", "-G")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
13 parameters['miniprot'] = value
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
14 def effective_params = merge_params(default_params, parameters, "miniprot")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
15 return effective_params
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
16 }
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
17
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
18 workflow miniprot {
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
19 take:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
20 fasta_genome_file //path: genome fasta file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
21 fasta_proteins_file //path: protein fasta file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
22 max_intron //int: max intron length
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
23 parameters // Map : extra parameter and parameter update
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
24 main:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
25 // println("Miniprot max intron: ${max_intron}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
26 def items_per_chunk = merge_params("-n 1000000000", parameters, "split_proteins").replaceFirst("-n ", "").toInteger()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
27 def protein_chunks
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
28 if (items_per_chunk == 1000000000) {
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
29 protein_chunks = fasta_proteins_file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
30 } else {
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
31 protein_chunks = split_proteins(fasta_proteins_file, items_per_chunk)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
32 }
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
33 run_miniprot(fasta_genome_file, protein_chunks.flatten(), max_intron, parameters)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
34
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
35 emit:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
36 miniprot_file = run_miniprot.out.miniprot_file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
37 }
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
38
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
39
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
40 process split_proteins {
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
41 input:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
42 path fasta_proteins_file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
43 val items_per_chunk
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
44 output:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
45 path 'output/*'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
46 script:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
47 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
48 #!/usr/bin/env python3
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
49 import os
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
50
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
51 os.makedirs("output", exist_ok=True)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
52 with open("${fasta_proteins_file}", 'rt') as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
53 items = 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
54 chunk = []
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
55 nextfile = 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
56 for line in f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
57 if line and line[0] == '>':
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
58 items += 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
59 if items >= ${items_per_chunk}:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
60 with open(f"output/{nextfile}.fa", "w") as outf:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
61 outf.write(''.join(chunk))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
62 chunk = []
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
63 nextfile += 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
64 items = 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
65 chunk.append(line)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
66 if chunk:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
67 with open(f"output/{nextfile}.fa", "w") as outf:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
68 outf.write(''.join(chunk))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
69 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
70 stub:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
71 print("items_per_chunk ${items_per_chunk}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
72 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
73 mkdir -p output
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
74 touch output/1.fa
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
75 touch output/2.fa
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
76 touch output/3.fa
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
77 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
78 }
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
79
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
80
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
81 process run_miniprot {
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
82 label 'huge_job'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
83 label 'long_job'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
84 input:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
85 path fasta_genome_file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
86 path fasta_proteins_file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
87 val max_intron
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
88 val parameters
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
89 output:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
90 path ('output/*.paf'), emit: 'miniprot_file'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
91
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
92 script:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
93 def paf_name = fasta_proteins_file.baseName.toString() + ".paf"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
94 def effective_params = get_effective_params(parameters, max_intron)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
95 // println("Miniprot params: ${effective_params}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
96 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
97 mkdir -p output
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
98 miniprot ${effective_params} ${fasta_genome_file} ${fasta_proteins_file} > output/${paf_name}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
99 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
100 stub:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
101 def paf_name = fasta_proteins_file.baseName.toString() + ".paf"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
102 def effective_params = get_effective_params(parameters, max_intron)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
103 println("Miniprot params: ${effective_params}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
104 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
105 mkdir -p output
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
106 touch output/${paf_name}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
107 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
108 }