Mercurial > repos > fubar > egapx_runner

annotate nf/subworkflows/ncbi/setup/main.nf @ 5:6effccc966d0 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 9e59da535540cb4d5c1c412bb2b0969744dfb0b0
author fubar
date Sun, 04 Aug 2024 01:59:37 +0000
parents d9c5c5b87fec
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 nextflow.enable.dsl=2
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
3
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
4 include { merge_params } from '../utilities'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
5
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
6 workflow setup_genome {
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
7 take:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
8 genome
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
9 organelles
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
10 parameters // Map : extra parameter and parameter update
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
11 main:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
12 get_genome_info(genome, organelles, parameters)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
13 emit:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
14 seqid_list = get_genome_info.out.seqid_list
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
15 gencoll_asn = get_genome_info.out.gencoll_asn
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
16 unpacked_genome = get_genome_info.out.fasta
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
17 genome_asn = get_genome_info.out.genome_asn
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
18 genome_asnb = get_genome_info.out.genome_asnb
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
19 max_intron = get_genome_info.out.max_intron
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
20 }
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
21
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
22
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
23 process get_genome_info {
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
24 debug true
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
25 input:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
26 path fasta_genome_file, stageAs: 'src/*'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
27 path organelles
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
28 val parameters
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
29 output:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
30 path '*.seqids', emit: 'seqid_list'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
31 path '*.asn', emit: 'gencoll_asn'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
32 path "${out_fasta}", emit: 'fasta'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
33 path "${genome_asn}", emit: 'genome_asn'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
34 path "${genome_asnb}", emit: 'genome_asnb'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
35 env max_intron, emit: 'max_intron'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
36 script:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
37 need_zcat = fasta_genome_file.toString().endsWith('.gz')
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
38 base_name_stripped = fasta_genome_file.baseName.toString().replaceAll(/\.(fa(sta)?|fna)(\.gz)?$/, "")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
39 indexed_fasta_name = fasta_genome_file.baseName.toString().replaceFirst(/\.(fa(sta)?|fna)(\.gz)?$/, ".fasta")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
40 if (! indexed_fasta_name.endsWith(".fasta")) {
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
41 indexed_fasta_name += ".fasta"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
42 }
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
43 genome_dir = "genome"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
44 fasta_dir = "fasta"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
45 out_fasta = fasta_dir + "/" + indexed_fasta_name
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
46 genome_asn = genome_dir + "/" + base_name_stripped + ".asn"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
47 genome_asnb = genome_dir + "/" + base_name_stripped + ".asnb"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
48 max_intron = parameters.max_intron
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
49 genome_size_threshold = parameters.genome_size_threshold
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 # echo "need_zcat: ${need_zcat}, out_fasta: ${out_fasta}"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
52 mkdir -p ${genome_dir}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
53 mkdir -p ${fasta_dir}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
54 if [[ ${need_zcat} == true ]]; then
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
55 # zcat ${fasta_genome_file} | sed 's/^\\(>gi|[0-9]\\+\\)|\\?\\([^ ]\\+\\)\\(.*\\)/\\1\\3/' > ${out_fasta}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
56 # zcat ${fasta_genome_file} > ${out_fasta}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
57 zcat ${fasta_genome_file} | sed 's/>\\([^ |]\\+\\)\\( .*\\)\\?\$/>lcl\\|\\1\\2/' > ${out_fasta}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
58 else
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
59 # sed 's/^\\(>gi|[0-9]\\+\\)|\\?\\([^ ]\\+\\)\\(.*\\)/\\1\\3/' ${fasta_genome_file} > ${out_fasta}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
60 # mv ${fasta_genome_file} ${out_fasta}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
61 sed 's/>\\([^ |]\\+\\)\\( .*\\)\\?\$/>lcl\\|\\1\\2/' ${fasta_genome_file} > ${out_fasta}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
62 fi
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
63 # Old way, now use gc_get_molecules. For multipart ids with gi first use the second part
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
64 # grep -oP "^>\\K[^ ]+" ${out_fasta} | sed 's/^\\(gi|[0-9]\\+\\)|\\([^|]\\+|[^|]\\+\\)|\\?/\\2/' >list.seqids
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
65 multireader -flags ParseRawID -out-format asn_text -input ${out_fasta} -output ${genome_asn}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
66 multireader -flags ParseRawID -out-format asn_binary -input ${out_fasta} -output ${genome_asnb}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
67 lds2_indexer -source ${genome_dir}/ -db LDS2
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
68 # Using all parts of multipart ids is preferrable, but slower - one more pass over genomic FASTA
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
69 gc_create -unplaced ${out_fasta} -unplaced-fmt fasta -fasta-parse-raw-id -gc-assm-name "EGAPx Test Assembly" -nogenbank -lds2 LDS2 >gencoll.asn
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
70 gc_get_molecules -gc-assembly gencoll.asn -filter all -level top-level > list.seqids
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
71
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
72 #TODO: subtract organelles from list
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
73
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
74 # This is a rough estimate because we don't need the more accurate size
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
75 genome_size=`wc -c <${out_fasta}`
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
76 # Max intron logic
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
77 if [ $genome_size_threshold -gt 0 ] && [ \$genome_size -lt $genome_size_threshold ]; then
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
78 # scale max intron to genome size, rounding up to nearest 100kb
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
79 (( max_intron = ($max_intron * genome_size / $genome_size_threshold + 99999) / 100000 * 100000 ))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
80 # echo "Setting max_intron to \$max_intron"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
81 else
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
82 max_intron=$max_intron
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
83 fi
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
84 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
85
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
86 stub:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
87 base_name_stripped = fasta_genome_file.baseName.toString().replaceAll(/\.(fa(sta)?|fna)(\.gz)?$/, "")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
88 indexed_fasta_name = fasta_genome_file.baseName.toString().replaceFirst(/\.(fa(sta)?|fna)(\.gz)?$/, ".fasta")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
89 if (! indexed_fasta_name.endsWith(".fasta")) {
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
90 indexed_fasta_name += ".fasta"
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 genome_dir = "genome"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
93 fasta_dir = "fasta"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
94 out_fasta = fasta_dir + "/" + indexed_fasta_name
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
95 genome_asn = genome_dir + "/" + base_name_stripped + ".asn"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
96 genome_asnb = genome_dir + "/" + base_name_stripped + ".asnb"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
97 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
98 mkdir -p $genome_dir
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
99 mkdir -p $fasta_dir
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
100 touch $out_fasta
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
101 touch $genome_asn
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
102 touch $genome_asnb
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
103 touch gencoll.asn
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
104 touch list.seqids
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
105 max_intron=10000
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
106 echo "Processing genome $fasta_genome_file"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
107 echo "Setting max_intron to \$max_intron"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
108 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
109 }
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
110
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
111
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
112 workflow setup_proteins {
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
113 take:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
114 proteins
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
115 parameters // Map : extra parameter and parameter update
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
116 main:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
117 convert_proteins(proteins)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
118 emit:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
119 unpacked_proteins = convert_proteins.out.unpacked_proteins
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
120 proteins_asn = convert_proteins.out.proteins_asn
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
121 proteins_asnb = convert_proteins.out.proteins_asnb
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
122 }
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
123
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
124
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
125 process convert_proteins {
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
126 input:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
127 path fasta_proteins_file, stageAs: 'src/*'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
128 output:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
129 path out_fasta, emit: 'unpacked_proteins'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
130 path proteins_asn, emit: 'proteins_asn'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
131 path proteins_asnb, emit: 'proteins_asnb'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
132 script:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
133 need_zcat = fasta_proteins_file.toString().endsWith('.gz')
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
134 base_name_stripped = fasta_proteins_file.baseName.toString().replaceAll(/\.(fa(sta)?|faa)(\.gz)?$/, "")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
135 fasta_name = base_name_stripped + ".faa"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
136
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
137 asn_dir = "asn"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
138 fasta_dir = "fasta"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
139 out_fasta = fasta_dir + "/" + fasta_name
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
140 proteins_asn = asn_dir + "/" + base_name_stripped + ".asn"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
141 proteins_asnb = asn_dir + "/" + base_name_stripped + ".asnb"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
142 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
143 mkdir -p ${asn_dir}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
144 mkdir -p ${fasta_dir}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
145 if [[ ${need_zcat} == true ]]; then
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
146 zcat ${fasta_proteins_file} | sed 's/>\\([^ |]\\+\\)\\( .*\\)\\?\$/>lcl\\|\\1\\2/' > ${out_fasta}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
147 else
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
148 sed 's/>\\([^ |]\\+\\)\\( .*\\)\\?\$/>lcl\\|\\1\\2/' ${fasta_proteins_file} > ${out_fasta}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
149 fi
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
150 multireader -flags ParseRawID -out-format asn_text -input ${out_fasta} -output ${proteins_asn}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
151 multireader -flags ParseRawID -out-format asn_binary -input ${out_fasta} -output ${proteins_asnb}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
152 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
153
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
154 stub:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
155 base_name_stripped = fasta_proteins_file.baseName.toString().replaceAll(/\.(fa(sta)?|faa)(\.gz)?$/, "")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
156 fasta_name = base_name_stripped + ".faa"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
157 asn_dir = "asn"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
158 fasta_dir = "fasta"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
159 out_fasta = fasta_dir + "/" + fasta_name
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
160 proteins_asn = asn_dir + "/" + base_name_stripped + ".asn"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
161 proteins_asnb = asn_dir + "/" + base_name_stripped + ".asnb"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
162
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
163 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
164 mkdir -p $asn_dir
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
165 mkdir -p $fasta_dir
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
166 touch $out_fasta
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
167 touch $proteins_asn
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
168 touch $proteins_asnb
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
169 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
170
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
171 }