Mercurial > repos > iuc > lexicmap

comparison lexicmap.xml @ 3:73480ee77517 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/lexicmap commit a9227295a4cebc34b17def7b5ca3e4506222b963
author iuc
date Fri, 26 Sep 2025 20:46:49 +0000
parents 73accf8f68a9
children 0004a511baff
comparison
equal deleted inserted replaced
2:73accf8f68a9 3:73480ee77517
5 </macros> 5 </macros>
6 <expand macro="bio_tools"/> 6 <expand macro="bio_tools"/>
7 <expand macro="requirements"/> 7 <expand macro="requirements"/>
8 8
9 <command detect_errors="exit_code"><![CDATA[ 9 <command detect_errors="exit_code"><![CDATA[
10 10 #if $db_opts.db_opts_selector == "histdb"
11 lexicmap search 11 #set INDICES = [db.extra_files_path for db in $db_opts.histdb]
12 12 #else:
13 --threads "\${GALAXY_SLOTS:-1}" 13 #set INDICES = $db_opts.lexicmap_index.fields.path.split(",")
14 14 #end if
15 ${load_whole_seeds} 15
16 ${all} 16 extract_query_ids() {
17 17 local input_files="\$1";
18 #if $db_opts.db_opts_selector == "histdb" 18 local query_ids="";
19 --index '${db_opts.histdb.extra_files_path}' 19 IFS=',' read -ra files <<< "\$input_files";
20 #else: 20 query_ids="";
21 --index '${db_opts.lexicmap_index.fields.path}' 21 for query_file in "\${files[@]}"; do
22 #end if 22 if file --mime-type "\$query_file" | grep -q "gzip"; then
23 23 query_ids+=\$(zcat "\$query_file" | grep '^>' | while IFS= read -r line; do clean="\${line#>}"; echo "\${clean%% *}>"; done);
24 #for $q in $query 24 else
25 '$q' 25 query_ids+=\$(cat "\$query_file" | grep '^>' | while IFS= read -r line; do clean="\${line#>}"; echo "\${clean%% *}>"; done);
26 #end for 26 fi
27 27 done;
28 --out-file '$out_file' 28 declare -g -a query_array=();
29 29 IFS='>' read -r -a query_array <<< "\$query_ids";
30 --top-n-genomes '$top_n_genomes' 30 }
31 31 &&
32 --align-band '$align_band' 32 #for $counter, $index in enumerate($INDICES):
33 --align-ext-len '$align_ext_len' 33 lexicmap search
34 --align-max-gap '$align_max_gap' 34
35 --align-min-match-len '$align_min_match_len' 35 --threads "\${GALAXY_SLOTS:-1}"
36 --align-min-match-pident '$align_min_match_pident' 36
37 --max-evalue '$max_evalue' 37 ${load_whole_seeds}
38 --max-query-conc '$max_query_conc' 38 ${all}
39 --seed-max-dist '$seed_max_dist' 39
40 --seed-max-gap '$seed_max_gap' 40 --index '${index}'
41 --seed-min-prefix '$seed_min_prefix' 41
42 --seed-min-single-prefix '$seed_min_single_prefix' 42 #for $q in $query
43 43 '$q'
44 #if $min_qcov_per_genome 44 #end for
45 --min-qcov-per-genome '$min_qcov_per_genome' 45
46 #end if 46 --out-file 'lexicmap_search_result__index${counter}.tsv'
47 47
48 #if $min_qcov_per_hsp 48 --top-n-genomes '$top_n_genomes'
49 --min-qcov-per-hsp '$min_qcov_per_hsp' 49
50 #end if 50 --align-band '$align_band'
51 51 --align-ext-len '$align_ext_len'
52 --align-max-gap '$align_max_gap'
53 --align-min-match-len '$align_min_match_len'
54 --align-min-match-pident '$align_min_match_pident'
55 --max-evalue '$max_evalue'
56 --max-query-conc '$max_query_conc'
57 --seed-max-dist '$seed_max_dist'
58 --seed-max-gap '$seed_max_gap'
59 --seed-min-prefix '$seed_min_prefix'
60 --seed-min-single-prefix '$seed_min_single_prefix'
61
62 #if $min_qcov_per_genome
63 --min-qcov-per-genome '$min_qcov_per_genome'
64 #end if
65
66 #if $min_qcov_per_hsp
67 --min-qcov-per-hsp '$min_qcov_per_hsp'
68 #end if
69 &&
70 #end for
71
72 #if len($INDICES) > 1
73 counter=0 &&
74 extract_query_ids '$query' &&
75 for ((i=0; i<\${#query_array[@]}; i++)); do
76 counter=\$((counter + 1));
77 lexicmap utils merge-search-results
78 --out-file "combined_result.\${counter}.tsv"
79 -q "\${query_array[\$i]}" lexicmap_search_result__index*.tsv
80 -j "\${GALAXY_SLOTS:-1}";
81 done &&
82 cat combined_result.*.tsv | awk 'NR==1 || $0 !~ /^query\tqlen\thits/' > '$out_file'
83 #else
84 mv lexicmap_search_result__index0.tsv '$out_file'
85 #end if
52 ]]></command> 86 ]]></command>
53 <inputs> 87 <inputs>
54 <param name="query" type="data" format="fasta.gz" label="LexicMap query file" multiple="true" help=""/> 88 <param name="query" type="data" format="fasta.gz" label="LexicMap query file" multiple="true" help=""/>
55 <conditional name="db_opts"> 89 <conditional name="db_opts">
56 <param name="db_opts_selector" type="select" label="LexiMap index source"> 90 <param name="db_opts_selector" type="select" label="LexiMap index source">
57 <option value="histdb" selected="true">From your history</option> 91 <option value="histdb" selected="true">From your history</option>
58 <option value="db">Locally installed LexicMap indexes</option> 92 <option value="db">Locally installed LexicMap indexes</option>
59 </param> 93 </param>
60 <when value="histdb"> 94 <when value="histdb">
61 <param name="histdb" type="data" format="lexicmap_index" optional="false" label="LexicMap index" /> 95 <param name="histdb" type="data" format="lexicmap_index" optional="false" multiple="true" label="LexicMap index" />
62 </when> 96 </when>
63 <when value="db"> 97 <when value="db">
64 <param name="lexicmap_index" type="select" optional="false" label="LexicMap index file"> 98 <param name="lexicmap_index" type="select" optional="false" multiple="true" label="LexicMap index file">
65 <options from_data_table="lexicmap_index"/> 99 <options from_data_table="lexicmap_index"/>
66 </param> 100 </param>
67 </when> 101 </when>
68 </conditional> 102 </conditional>
69 <param argument="--top-n-genomes" type="integer" value="0" label="Keep top N genome matches for a query (0 for all)" /> 103 <param argument="--top-n-genomes" type="integer" value="0" label="Keep top N genome matches for a query (0 for all)" />
98 </conditional> 132 </conditional>
99 </actions> 133 </actions>
100 </data> 134 </data>
101 </outputs> 135 </outputs>
102 <tests> 136 <tests>
103 <!-- Test 1 - query a local index with one query --> 137 <!-- Test 1 - query one local index with one query -->
104 <test expect_num_outputs="1"> 138 <test expect_num_outputs="1">
105 <conditional name="db_opts"> 139 <conditional name="db_opts">
106 <param name="db_opts_selector" value="db"/> 140 <param name="db_opts_selector" value="db"/>
107 <param name="lexicmap_index" value="LexicMapIndex1" /> 141 <param name="lexicmap_index" value="LexicMapIndex1" />
108 </conditional> 142 </conditional>
110 <section name="advanced_settings"> 144 <section name="advanced_settings">
111 <param name="load_whole_seeds" value="true" /> 145 <param name="load_whole_seeds" value="true" />
112 </section> 146 </section>
113 <output name="out_file" value="lexicmap_query_result.tsv" /> 147 <output name="out_file" value="lexicmap_query_result.tsv" />
114 </test> 148 </test>
115 <!-- Test 2 - query a local index with multiple query files --> 149 <!-- Test 2 - query one local index with multiple query files -->
116 <test expect_num_outputs="1"> 150 <test expect_num_outputs="1">
117 <conditional name="db_opts"> 151 <conditional name="db_opts">
118 <param name="db_opts_selector" value="db"/> 152 <param name="db_opts_selector" value="db"/>
119 <param name="lexicmap_index" value="LexicMapIndex1" /> 153 <param name="lexicmap_index" value="LexicMapIndex1" />
120 </conditional> 154 </conditional>
122 <section name="advanced_settings"> 156 <section name="advanced_settings">
123 <param name="load_whole_seeds" value="true" /> 157 <param name="load_whole_seeds" value="true" />
124 </section> 158 </section>
125 <output name="out_file" value="lexicmap_query_result2.tsv" /> 159 <output name="out_file" value="lexicmap_query_result2.tsv" />
126 </test> 160 </test>
127 <!-- Test 3 - query a index found in the history with one query --> 161 <!-- Test 3 - query two local index with one query file -->
162 <test expect_num_outputs="1">
163 <conditional name="db_opts">
164 <param name="db_opts_selector" value="db"/>
165 <param name="lexicmap_index" value="LexicMapIndexCombined" />
166 </conditional>
167 <param name="query" value="lexicmap_query.fasta.gz" />
168 <section name="advanced_settings">
169 <param name="load_whole_seeds" value="true" />
170 </section>
171 <output name="out_file" value="lexicmap_query_result.tsv" />
172 </test>
173 <!-- Test 4 - query two local index with multiple query files -->
174 <test expect_num_outputs="1">
175 <conditional name="db_opts">
176 <param name="db_opts_selector" value="db"/>
177 <param name="lexicmap_index" value="LexicMapIndexCombined" />
178 </conditional>
179 <param name="query" value="lexicmap_query.fasta.gz,lexicmap_query2.fasta.gz,lexicmap_query3.fasta" />
180 <section name="advanced_settings">
181 <param name="load_whole_seeds" value="true" />
182 </section>
183 <output name="out_file" value="lexicmap_query_result4.tsv" />
184 </test>
185 <!-- Test 5 - query one local index with multiple query files, where only one query will get hits -->
186 <test expect_num_outputs="1">
187 <conditional name="db_opts">
188 <param name="db_opts_selector" value="db"/>
189 <param name="lexicmap_index" value="LexicMapIndex2" />
190 </conditional>
191 <param name="query" value="lexicmap_query.fasta.gz,lexicmap_query2.fasta.gz,lexicmap_query3.fasta" />
192 <section name="advanced_settings">
193 <param name="load_whole_seeds" value="true" />
194 </section>
195 <output name="out_file" value="lexicmap_query_result3.tsv" />
196 </test>
197 <!-- Test 6 - query multiple local index with multiple query files -->
198 <test expect_num_outputs="1">
199 <conditional name="db_opts">
200 <param name="db_opts_selector" value="db"/>
201 <param name="lexicmap_index" value="LexicMapIndex1,LexicMapIndex2,LexicMapIndexCombined" />
202 </conditional>
203
204 <param name="query" value="lexicmap_query.fasta.gz,lexicmap_query2.fasta.gz,lexicmap_query3.fasta" />
205 <section name="advanced_settings">
206 <param name="load_whole_seeds" value="true" />
207 </section>
208 <output name="out_file" value="lexicmap_query_result6.tsv" />
209 </test>
210 <!-- Test 7 - query one index found in the history with one query -->
128 <test expect_num_outputs="1"> 211 <test expect_num_outputs="1">
129 <conditional name="db_opts"> 212 <conditional name="db_opts">
130 <param name="db_opts_selector" value="histdb"/> 213 <param name="db_opts_selector" value="histdb"/>
131 <param name="histdb" ftype="lexicmap_index" class="Directory" value="db.lmi" /> 214 <param name="histdb" ftype="lexicmap_index" class="Directory" value="db.lmi" />
132 </conditional> 215 </conditional>
134 <param name="query" value="lexicmap_query.fasta.gz" /> 217 <param name="query" value="lexicmap_query.fasta.gz" />
135 <section name="advanced_settings"> 218 <section name="advanced_settings">
136 <param name="load_whole_seeds" value="true" /> 219 <param name="load_whole_seeds" value="true" />
137 </section> 220 </section>
138 <output name="out_file" value="lexicmap_query_result.tsv" /> 221 <output name="out_file" value="lexicmap_query_result.tsv" />
222 </test>
223 <!-- Test 8 - query two index found in the history with one query -->
224 <test expect_num_outputs="1">
225 <conditional name="db_opts">
226 <param name="db_opts_selector" value="histdb"/>
227 <param name="histdb" ftype="lexicmap_index" class="Directory" value="db.lmi,db2.lmi" />
228 </conditional>
229 <param name="top_n_genomes" value="0" />
230 <param name="query" value="lexicmap_query.fasta.gz,lexicmap_query3.fasta" />
231 <section name="advanced_settings">
232 <param name="load_whole_seeds" value="true" />
233 </section>
234 <output name="out_file" value="lexicmap_query_result5.tsv" />
139 </test> 235 </test>
140 </tests> 236 </tests>
141 <help><![CDATA[ 237 <help><![CDATA[
142 238
143 Search sequences against an LexicMap index Database. For more information about settings 239 Search sequences against an LexicMap index Database. For more information about settings
170 21. cigar, CIGAR string of the alignment. (optional with --all) 266 21. cigar, CIGAR string of the alignment. (optional with --all)
171 22. qseq, Aligned part of query sequence. (optional with --all) 267 22. qseq, Aligned part of query sequence. (optional with --all)
172 23. sseq, Aligned part of subject sequence. (optional with --all) 268 23. sseq, Aligned part of subject sequence. (optional with --all)
173 24. align, Alignment text ("|" and " ") between qseq and sseq. (optional with --all) 269 24. align, Alignment text ("|" and " ") between qseq and sseq. (optional with --all)
174 270
271 When running against multiple indices lexicmap utils merge-search-results will be used to
272 merge the search results. For more information please visit:
273 https://bioinf.shenwei.me/LexicMap/usage/utils/merge-search-results/
274
275 Note: if the query id contains spaces, only the first part (before the first space) will be kept as the query id.
175 @info@ 276 @info@
176 ]]></help> 277 ]]></help>
177 <expand macro="citations" /> 278 <expand macro="citations" />
178 </tool> 279 </tool>