Mercurial > repos > iuc > data_manager_build_kraken2_database
comparison data_manager/kraken2_build_database.xml @ 13:150ea60a4ebc draft default tip
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/data_managers/data_manager_build_kraken2_database commit a108f20aebc04574a8bd0a90b955064439a50852
| author | iuc |
|---|---|
| date | Wed, 05 Nov 2025 13:31:52 +0000 |
| parents | bf7f964a12b7 |
| children |
comparison
equal
deleted
inserted
replaced
| 12:bf7f964a12b7 | 13:150ea60a4ebc |
|---|---|
| 1 <tool id="kraken2_build_database" name="Kraken2" tool_type="manage_data" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@"> | 1 <tool id="kraken2_build_database" name="Kraken2" tool_type="manage_data" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@"> |
| 2 <description>database builder</description> | 2 <description>database builder</description> |
| 3 <macros> | 3 <macros> |
| 4 <token name="@TOOL_VERSION@">2.1.3</token> | 4 <token name="@TOOL_VERSION@">2.1.6</token> |
| 5 <token name="@VERSION_SUFFIX@">6</token> | 5 <token name="@VERSION_SUFFIX@">0</token> |
| 6 <token name="@PROFILE@">22.01</token> | 6 <token name="@PROFILE@">24.0</token> |
| 7 <xml name="common_params"> | 7 <xml name="common_params"> |
| 8 <param name="kmer_len" type="integer" value="35" label="K-mer length in BP" /> | 8 <param name="kmer_len" type="integer" value="35" label="K-mer length in BP" /> |
| 9 <param name="minimizer_len" type="integer" value="31" label="Minimizer length" /> | 9 <param name="minimizer_len" type="integer" value="31" label="Minimizer length" /> |
| 10 <param name="minimizer_spaces" type="integer" value="7" label="Minimizer spaces" /> | 10 <param name="minimizer_spaces" type="integer" value="7" label="Minimizer spaces" /> |
| 11 <param name="load_factor" type="float" value="0.7" min="0" max="1" label="Load factor" help="Proportion of the hash table to be populated" /> | 11 <param name="load_factor" type="float" value="0.7" min="0" max="1" label="Load factor" help="Proportion of the hash table to be populated" /> |
| 12 <param name="clean" type="boolean" truevalue="--clean" falsevalue="" checked="true" label="Clean up extra files" /> | 12 <param name="clean" type="boolean" truevalue="--clean" falsevalue="" checked="false" label="Clean up extra files. Note: If the extra files are removed this DB cannot be used to build a bracken DB!" /> |
| 13 </xml> | 13 </xml> |
| 14 <xml name="viral"> | 14 <xml name="viral"> |
| 15 <option value="viral">Viral (viral; ~0.5 GB)</option> | 15 <option value="viral">Viral (viral; ~0.5 GB)</option> |
| 16 </xml> | 16 </xml> |
| 17 <xml name="minusb"> | 17 <xml name="minusb"> |
| 41 <xml name="pluspfp_08gb"> | 41 <xml name="pluspfp_08gb"> |
| 42 <option value="pluspfp_08gb">PlusPFP-8 (PlusPFP with DB capped at 8 GB; ~7.5 GB)</option> | 42 <option value="pluspfp_08gb">PlusPFP-8 (PlusPFP with DB capped at 8 GB; ~7.5 GB)</option> |
| 43 </xml> | 43 </xml> |
| 44 <xml name="pluspfp_16gb"> | 44 <xml name="pluspfp_16gb"> |
| 45 <option value="pluspfp_16gb">PlusPFP-16 (PlusPFP with DB capped at 16 GB; ~15 GB)</option> | 45 <option value="pluspfp_16gb">PlusPFP-16 (PlusPFP with DB capped at 16 GB; ~15 GB)</option> |
| 46 </xml> | |
| 47 <xml name="core_nt"> | |
| 48 <option value="core_nt">core_nt (Very large collection, inclusive of GenBank, RefSeq, TPA and PDB; ~182 GB)</option> | |
| 49 </xml> | 46 </xml> |
| 50 </macros> | 47 </macros> |
| 51 <xrefs> | 48 <xrefs> |
| 52 <xref type="bio.tools">kraken2</xref> | 49 <xref type="bio.tools">kraken2</xref> |
| 53 </xrefs> | 50 </xrefs> |
| 54 <requirements> | 51 <requirements> |
| 55 <requirement type="package" version="@TOOL_VERSION@">kraken2</requirement> | 52 <requirement type="package" version="@TOOL_VERSION@">kraken2</requirement> |
| 56 <requirement type="package" version="3.13">python</requirement> | |
| 57 </requirements> | 53 </requirements> |
| 58 <version_command>kraken2 -version | head -n 1 | awk '{print $NF}'</version_command> | 54 <version_command>kraken2 -version | head -n 1 | awk '{print $NF}'</version_command> |
| 59 <command detect_errors="exit_code"><![CDATA[ | 55 <command detect_errors="exit_code"><![CDATA[ |
| 60 python '$__tool_directory__/kraken2_build_database.py' | 56 #import datetime |
| 61 '$out_file' | 57 #import re |
| 62 --database-type '$database_type.database_type' | 58 |
| 63 #if $database_type.database_type == "standard_local_build" | 59 #set now = datetime.datetime.utcnow().strftime("%Y-%m-%dT%H%M%SZ") |
| 64 --threads \${GALAXY_SLOTS:-1} | 60 #set commands = [] |
| 65 --kmer-len $database_type.kmer_len | 61 mkdir '$out_file.extra_files_path' && |
| 66 --minimizer-len $database_type.minimizer_len | 62 |
| 67 --minimizer-spaces $database_type.minimizer_spaces | 63 #if $database_type.database_type == "standard_prebuilt" or $database_type.database_type == "special_prebuilt" or $database_type.database_type == "amplicon_prebuilt" |
| 68 --load-factor $database_type.load_factor | 64 #set prebuilt_name = { |
| 69 $database_type.clean | 65 'viral': "Viral", |
| 70 #else if $database_type.database_type == "standard_prebuilt" | 66 'minusb': "MinusB (archaea, viral, plasmid, human, UniVec_Core)", |
| 71 --prebuilt-db '$database_type.prebuild.prebuilt_db' | 67 'standard': "Standard-Full (archaea, bacteria, viral, plasmid, human,UniVec_Core)", |
| 72 --prebuilt-date '$database_type.prebuild.prebuilt_date' | 68 'standard_08gb': "Standard-8 (Standard with DB capped at 8 GB)", |
| 69 'standard_16gb': "Standard-16 (Standard with DB capped at 16 GB)", | |
| 70 'pluspf': "PlusPF (Standard plus protozoa and fungi)", | |
| 71 'pluspf_08gb': "PlusPF-8 (PlusPF with DB capped at 8 GB)", | |
| 72 'pluspf_16gb': "PlusPF-16 (PlusPF with DB capped at 16 GB)", | |
| 73 'pluspfp': "PlusPFP (Standard plus protozoa, fungi and plant)", | |
| 74 'pluspfp_08gb': "PlusPFP-8 (PlusPFP with DB capped at 8 GB)", | |
| 75 'pluspfp_16gb': "PlusPFP-16 (PlusPFP with DB capped at 16 GB)", | |
| 76 } | |
| 77 #set special_name = { | |
| 78 "core_nt_20250609": "Very large collection, inclusive of GenBank, RefSeq, TPA and PDB (July, 2025)", | |
| 79 "core_nt_20241228": "Very large collection, inclusive of GenBank, RefSeq, TPA and PDB (December, 2024)", | |
| 80 "core_nt_20240904": "Very large collection, inclusive of GenBank, RefSeq, TPA and PDB (September, 2024)", | |
| 81 "gtdb_genome_reps_20250609": "GTDB v226 (Bacterial and archaeal)", | |
| 82 "gtdb_genome_reps_20241109": "GTDB v220 (Bacterial and archaeal)", | |
| 83 "eupathdb48_20230407": "EuPathDB-46 (April 18, 2023)", | |
| 84 "eupathdb48_20201113": "EuPathDB-46 (November 13, 2020)" | |
| 85 } | |
| 86 #set amplicon_name = { | |
| 87 "16S_Greengenes13.5_20200326": "Greengenes 13.5", | |
| 88 "16S_RDP11.5_20200326": "RDP 11.5", | |
| 89 "16S_Silva132_20200326": "Silva 132", | |
| 90 "16S_Silva138_20200326": "Silva 138" | |
| 91 } | |
| 92 | |
| 93 #set date_url_str = str($database_type.prebuilt.prebuilt_date).replace('-', '') | |
| 94 #set display_name = prebuilt_name.get(str($database_type.prebuilt.prebuilt_db)) | |
| 95 #if not display_name | |
| 96 #set display_name = special_name.get(str($database_type.prebuilt.prebuilt_db)+"_"+date_url_str) | |
| 97 #end if | |
| 98 #if not display_name | |
| 99 #set display_name = amplicon_name.get(str($database_type.prebuilt.prebuilt_db)+"_"+date_url_str) | |
| 100 #end if | |
| 101 | |
| 102 #set database_value = "_".join([now, "standard_prebuilt", str($database_type.prebuilt.prebuilt_db), str($database_type.prebuilt.prebuilt_date)]) | |
| 103 #set database_name = " ".join(["Prebuilt Refseq indexes: ", display_name, "(Version: ", str($database_type.prebuilt.prebuilt_date), "- Downloaded:", now + ")"]) | |
| 104 | |
| 105 ## the 16S dbs have a different link and file name | |
| 106 ## and are stored in a subfolder | |
| 107 #if $database_type.database_type == "amplicon_prebuilt" | |
| 108 #silent commands.append("wget https://genome-idx.s3.amazonaws.com/kraken/" + str($database_type.prebuilt.prebuilt_db) + "_" + date_url_str + ".tgz") | |
| 109 #silent commands.append("mkdir -p '" + $out_file.extra_files_path + "/" + database_value + "'/tmp_extract") | |
| 110 #silent commands.append("tar -xzf " + str($database_type.prebuilt.prebuilt_db) + "_" + date_url_str + ".tgz -C '" + $out_file.extra_files_path + "/" + database_value + "'/tmp_extract") | |
| 111 #silent commands.append("topdir=$(find '" + $out_file.extra_files_path + "/" + database_value + "/tmp_extract' -mindepth 1 -maxdepth 1 -type d | head -n 1)") | |
| 112 #silent commands.append("if [ -n \"$topdir\" ]") | |
| 113 #silent commands.append("then") | |
| 114 #silent commands.append(" mv \"$topdir\"/* '" + $out_file.extra_files_path + "/" + database_value + "/'") | |
| 115 #silent commands.append("fi") | |
| 116 #silent commands.append("rm -rf '" + $out_file.extra_files_path + "/" + database_value + "/tmp_extract'") | |
| 117 #else | |
| 118 #silent commands.append("wget https://genome-idx.s3.amazonaws.com/kraken/k2_" + str($database_type.prebuilt.prebuilt_db) + "_" + date_url_str + ".tar.gz") | |
| 119 #silent commands.append("mkdir -p '" + $out_file.extra_files_path + "/" + database_value + "'") | |
| 120 #silent commands.append("tar -xzf k2_" + str($database_type.prebuilt.prebuilt_db) + "_" + date_url_str + ".tar.gz -C '" + $out_file.extra_files_path + "/" + database_value + "'") | |
| 121 #end if | |
| 122 | |
| 73 #else if $database_type.database_type == "minikraken" | 123 #else if $database_type.database_type == "minikraken" |
| 74 --minikraken2-version '$database_type.minikraken2_version' | 124 #set database_value = "_".join([now, "minikraken2", str($database_type.minikraken2_version), "8GB"]) |
| 75 #else if $database_type.database_type == "special_prebuilt" | 125 #set database_name = " ".join(["Minikraken2", str($database_type.minikraken2_version), "(Created:", now + ")"]) |
| 76 --prebuilt-db '$database_type.special_prebuild.prebuilt_db' | 126 |
| 77 --prebuilt-date '$database_type.special_prebuild.prebuilt_date' | 127 #silent commands.append("wget 'https://genome-idx.s3.amazonaws.com/kraken/minikraken2_" + str($database_type.minikraken2_version) + "_8GB_201904.tgz'") |
| 78 #else if $database_type.database_type == "special" | 128 #silent commands.append("mkdir -p '" + $out_file.extra_files_path + "'/'" + database_value + "'") |
| 79 --threads \${GALAXY_SLOTS:-1} | 129 #silent commands.append("tar -xzf 'minikraken2_" + str($database_type.minikraken2_version) + "_8GB_201904.tgz' -C '" + $out_file.extra_files_path + "'/'" + database_value + "'") |
| 80 --special-database-type '$database_type.special_database_type' | 130 #else |
| 81 --kmer-len $database_type.kmer_len | 131 #if $database_type.database_type == "standard_local_build" |
| 82 --minimizer-len $database_type.minimizer_len | 132 #set database_value = "_".join([now, "standard", "kmer-len", str($database_type.kmer_len), "minimizer-len", str($database_type.minimizer_len), "minimizer-spaces", str($database_type.minimizer_spaces), "load-factor", str($database_type.load_factor)]) |
| 83 --minimizer-spaces $database_type.minimizer_spaces | 133 #set database_name = " ".join(["Standard (Local Build)", "Created:", now + ",", "kmer-len=" + str($database_type.kmer_len) + ",", "minimizer-len=" + str($database_type.minimizer_len) + ",", "minimizer-spaces=" + str($database_type.minimizer_spaces) + ",", "load-factor=" + str($database_type.load_factor) + ")"]) |
| 84 --load-factor $database_type.load_factor | 134 #else if $database_type.database_type == "special" |
| 85 $database_type.clean | 135 #set special_database_names = {"rdp": "RDP", "greengenes": "Greengenes", "silva": "Silva"} |
| 86 #else if $database_type.database_type == "custom" | 136 #set database_value = "_".join([now, str($database_type.special_database_type), "kmer-len", str($database_type.kmer_len), "minimizer-len", str($database_type.minimizer_len), "minimizer-spaces", str($database_type.minimizer_spaces), "load-factor", str($database_type.load_factor)]) |
| 87 --threads \${GALAXY_SLOTS:-1} | 137 #set database_name = " ".join([special_database_names[str($database_type.special_database_type)], "(Created:", now + ",", "kmer-len=" + str($database_type.kmer_len) + ",", "minimizer-len=" + str($database_type.minimizer_len) + ",", "minimizer-spaces=" + str($database_type.minimizer_spaces) + ",", "load-factor=" + str($database_type.load_factor) + ")"]) |
| 88 --custom-fasta '$database_type.custom_fasta' | 138 #else if $database_type.database_type == "custom" |
| 89 --custom-database-name '$database_type.custom_database_name' | 139 #set custom_database_name = re.sub(r'[^\w_.-]+', '_', str($database_type.custom_database_name)).strip('_') |
| 90 --custom-source-info '$database_type.custom_source_info' | 140 #set database_name = " ".join([custom_database_name, "(" + str($database_type.custom_source_info) + ",", "kmer-len=" + str($database_type.kmer_len) + ",", "minimizer-len=" + str($database_type.minimizer_len) + ",", "minimizer-spaces=" + str($database_type.minimizer_spaces) + ",", "load-factor=" + str($database_type.load_factor) + ")"]) |
| 91 $database_type.skip_maps | 141 #set database_value = "_".join([now, custom_database_name, "kmer-len", str($database_type.kmer_len), "minimizer-len", str($database_type.minimizer_len), "minimizer-spaces", str($database_type.minimizer_spaces), "load-factor", str($database_type.load_factor)]) |
| 92 --kmer-len $database_type.kmer_len | 142 #else |
| 93 --minimizer-len $database_type.minimizer_len | 143 >&2 echo "invalid database_type: $database_type.database_type" |
| 94 --minimizer-spaces $database_type.minimizer_spaces | 144 #end if |
| 95 --load-factor $database_type.load_factor | 145 |
| 96 $database_type.clean | 146 #if $database_type.database_type == "custom" |
| 147 #silent command = ["kraken2-build", "--threads", '"${GALAXY_SLOTS:-1}"', | |
| 148 "--download-taxonomy", | |
| 149 "--db", "'" + $out_file.extra_files_path + "'/'" + database_value + "'", | |
| 150 str($database_type.skip_maps)] | |
| 151 #silent commands.append(" ".join(command)) | |
| 152 #silent command = ["kraken2-build", "--threads", '"${GALAXY_SLOTS:-1}"', | |
| 153 "--add-to-library", "'" + str($database_type.custom_fasta) + "'", | |
| 154 "--db", "'" + $out_file.extra_files_path + "'/'" + database_value + "'"] | |
| 155 #silent commands.append(" ".join(command)) | |
| 156 #end if | |
| 157 | |
| 158 #silent command = ["kraken2-build", "--threads", '"${GALAXY_SLOTS:-1}"'] | |
| 159 #if $database_type.database_type == "standard_local_build" | |
| 160 #silent command.append("--standard") | |
| 161 #else if $database_type.database_type == "special" | |
| 162 #silent command.extend(["--special", str($database_type.special_database_type)]) | |
| 163 #else if $database_type.database_type == "custom" | |
| 164 #silent command.append("--build") | |
| 165 #end if | |
| 166 #silent command.extend([ | |
| 167 "--kmer-len", str($database_type.kmer_len), | |
| 168 "--minimizer-len", str($database_type.minimizer_len), | |
| 169 "--minimizer-spaces", str($database_type.minimizer_spaces), | |
| 170 "--load-factor", str($database_type.load_factor), | |
| 171 "--db", "'" + $out_file.extra_files_path + "'/'" + database_value + "'"]) | |
| 172 #silent commands.append(" ".join(command)) | |
| 173 | |
| 174 #if $database_type.clean | |
| 175 #silent command = ["kraken2-build", "--threads", '"${GALAXY_SLOTS:-1}"', | |
| 176 "--clean", | |
| 177 "--db", "'" + $out_file.extra_files_path + "'/'" + database_value + "'"] | |
| 178 #silent commands.append(" ".join(command)) | |
| 179 #end if | |
| 97 #end if | 180 #end if |
| 181 | |
| 182 #for command in commands | |
| 183 ## In test mode the tool executes `echo COMMAND` instead of `COMMAND` | |
| 184 #if $run_test_command == "false" | |
| 185 echo | |
| 186 #end if | |
| 187 #echo command | |
| 188 && | |
| 189 #end for | |
| 190 | |
| 191 echo '{"data_tables": {"kraken2_databases": [{"value": "$database_value", "name": "$database_name", "path": "$database_value"}]}}' > '$out_file' | |
| 98 ]]> | 192 ]]> |
| 99 </command> | 193 </command> |
| 100 <inputs> | 194 <inputs> |
| 101 <conditional name="database_type"> | 195 <conditional name="database_type"> |
| 102 <param name="database_type" type="select" multiple="false" label="Database Type"> | 196 <param name="database_type" type="select" multiple="false" label="Database Type"> |
| 103 <option value="standard_local_build">Standard, Local Build</option> | 197 <option value="standard_local_build">Standard, Local Build</option> |
| 104 <option value="standard_prebuilt">Pre-Built Refseq indexes</option> | 198 <option value="standard_prebuilt">Pre-Built Refseq indexes</option> |
| 105 <option value="minikraken">MiniKraken</option> | 199 <option value="minikraken">MiniKraken</option> |
| 106 <option value="special_prebuilt">Special Pre-Built indexes</option> | 200 <option value="special_prebuilt">Special Pre-Built indexes</option> |
| 201 <option value="amplicon_prebuilt">16S Pre-Built indexes</option> | |
| 107 <option value="special">Special</option> | 202 <option value="special">Special</option> |
| 108 <option value="custom">Custom</option> | 203 <option value="custom">Custom</option> |
| 109 </param> | 204 </param> |
| 110 <when value="standard_local_build"> | 205 <when value="standard_local_build"> |
| 111 <expand macro="common_params" /> | 206 <expand macro="common_params" /> |
| 112 </when> | 207 </when> |
| 113 <when value="standard_prebuilt"> | 208 <when value="standard_prebuilt"> |
| 114 <conditional name="prebuild"> | 209 <conditional name="prebuilt"> |
| 115 <param name="prebuilt_date" type="select" label="Select index build date"> | 210 <param name="prebuilt_date" type="select" label="Select index build date"> |
| 211 <option value="2025-07-14">July 14, 2025</option> | |
| 116 <option value="2024-12-28">December 28, 2024</option> | 212 <option value="2024-12-28">December 28, 2024</option> |
| 117 <option value="2024-09-04">September 4, 2024</option> | 213 <option value="2024-09-04">September 4, 2024</option> |
| 118 <option value="2024-06-05">June 5, 2024</option> | 214 <option value="2024-06-05">June 5, 2024</option> |
| 119 <option value="2024-01-12">January 12, 2024</option> | 215 <option value="2024-01-12">January 12, 2024</option> |
| 120 <option value="2023-06-05">June 5, 2023</option> | 216 <option value="2023-06-05">June 5, 2023</option> |
| 122 <option value="2021-05-17">May 17, 2021</option> | 218 <option value="2021-05-17">May 17, 2021</option> |
| 123 <option value="2021-01-27">January 27, 2021</option> | 219 <option value="2021-01-27">January 27, 2021</option> |
| 124 <option value="2020-12-02">December 2, 2020</option> | 220 <option value="2020-12-02">December 2, 2020</option> |
| 125 <option value="2020-09-19">September 19, 2020</option> | 221 <option value="2020-09-19">September 19, 2020</option> |
| 126 </param> | 222 </param> |
| 223 <when value="2025-07-14"> | |
| 224 <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download"> | |
| 225 <expand macro="viral"/> | |
| 226 <expand macro="minusb"/> | |
| 227 <expand macro="standard"/> | |
| 228 <expand macro="standard_08gb"/> | |
| 229 <expand macro="standard_16gb"/> | |
| 230 <expand macro="pluspf"/> | |
| 231 <expand macro="pluspf_08gb"/> | |
| 232 <expand macro="pluspf_16gb"/> | |
| 233 <expand macro="pluspfp"/> | |
| 234 <expand macro="pluspfp_08gb"/> | |
| 235 <expand macro="pluspfp_16gb"/> | |
| 236 </param> | |
| 237 </when> | |
| 127 <when value="2024-12-28"> | 238 <when value="2024-12-28"> |
| 128 <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download"> | 239 <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download"> |
| 129 <expand macro="viral"/> | 240 <expand macro="viral"/> |
| 130 <expand macro="minusb"/> | 241 <expand macro="minusb"/> |
| 131 <expand macro="standard"/> | 242 <expand macro="standard"/> |
| 135 <expand macro="pluspf_08gb"/> | 246 <expand macro="pluspf_08gb"/> |
| 136 <expand macro="pluspf_16gb"/> | 247 <expand macro="pluspf_16gb"/> |
| 137 <expand macro="pluspfp"/> | 248 <expand macro="pluspfp"/> |
| 138 <expand macro="pluspfp_08gb"/> | 249 <expand macro="pluspfp_08gb"/> |
| 139 <expand macro="pluspfp_16gb"/> | 250 <expand macro="pluspfp_16gb"/> |
| 140 <expand macro="core_nt"/> | |
| 141 </param> | 251 </param> |
| 142 </when> | 252 </when> |
| 143 <when value="2024-09-04"> | 253 <when value="2024-09-04"> |
| 144 <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download"> | 254 <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download"> |
| 145 <expand macro="viral"/> | 255 <expand macro="viral"/> |
| 150 <expand macro="pluspf"/> | 260 <expand macro="pluspf"/> |
| 151 <expand macro="pluspf_08gb"/> | 261 <expand macro="pluspf_08gb"/> |
| 152 <expand macro="pluspf_16gb"/> | 262 <expand macro="pluspf_16gb"/> |
| 153 <expand macro="pluspfp"/> | 263 <expand macro="pluspfp"/> |
| 154 <expand macro="pluspfp_08gb"/> | 264 <expand macro="pluspfp_08gb"/> |
| 155 <expand macro="core_nt"/> | |
| 156 </param> | 265 </param> |
| 157 </when> | 266 </when> |
| 158 <when value="2024-06-05"> | 267 <when value="2024-06-05"> |
| 159 <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download"> | 268 <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download"> |
| 160 <expand macro="viral"/> | 269 <expand macro="viral"/> |
| 261 </param> | 370 </param> |
| 262 </when> | 371 </when> |
| 263 </conditional> | 372 </conditional> |
| 264 </when> | 373 </when> |
| 265 <when value="special_prebuilt"> | 374 <when value="special_prebuilt"> |
| 266 <conditional name="special_prebuild"> | 375 <conditional name="prebuilt"> |
| 267 <param name="special_prebuilt_db" type="select" multiple="false" label="Select pre-built database to download"> | 376 <param name="xyz" type="select" multiple="false" label="Select pre-built database to download"> |
| 268 <option value="gtdb_genome_reps_20241109">GTDB v220 (Bacterial and archaeal; ~497 GB) (December 13, 2024)</option> | 377 <option value="core_nt_20250609">Very large collection, inclusive of GenBank, RefSeq, TPA and PDB (July, 2025)</option> |
| 378 <option value="core_nt_20241228">Very large collection, inclusive of GenBank, RefSeq, TPA and PDB (December, 2024)</option> | |
| 379 <option value="core_nt_20240904">Very large collection, inclusive of GenBank, RefSeq, TPA and PDB (September, 2024)</option> | |
| 380 <option value="gtdb_genome_reps_20250609">GTDB v226 (Bacterial and archaeal; ~497 GB) (July, 2025)</option> | |
| 381 <option value="gtdb_genome_reps_20241109">GTDB v220 (Bacterial and archaeal; ~644 GB) (December 13, 2024)</option> | |
| 269 <option value="eupathdb48_20230407">EuPathDB-46 (April 18, 2023)</option> | 382 <option value="eupathdb48_20230407">EuPathDB-46 (April 18, 2023)</option> |
| 270 <option value="eupathdb48_20201113">EuPathDB-46 (November 13, 2020)</option> | 383 <option value="eupathdb48_20201113">EuPathDB-46 (November 13, 2020)</option> |
| 271 </param> | 384 </param> |
| 385 <when value="core_nt_20250609"> | |
| 386 <param name="prebuilt_db" type="hidden" value="core_nt"/> | |
| 387 <param name="prebuilt_date" type="hidden" value="20250609"/> | |
| 388 </when> | |
| 389 <when value="core_nt_20241228"> | |
| 390 <param name="prebuilt_db" type="hidden" value="core_nt"/> | |
| 391 <param name="prebuilt_date" type="hidden" value="20241228"/> | |
| 392 </when> | |
| 393 <when value="core_nt_20240904"> | |
| 394 <param name="prebuilt_db" type="hidden" value="core_nt"/> | |
| 395 <param name="prebuilt_date" type="hidden" value="20240904"/> | |
| 396 </when> | |
| 397 <when value="gtdb_genome_reps_20250609"> | |
| 398 <param name="prebuilt_db" type="hidden" value="gtdb_genome_reps"/> | |
| 399 <param name="prebuilt_date" type="hidden" value="2025-06-09"/> | |
| 400 </when> | |
| 272 <when value="gtdb_genome_reps_20241109"> | 401 <when value="gtdb_genome_reps_20241109"> |
| 273 <param name="prebuilt_db" type="hidden" value="gtdb_genome_reps"/> | 402 <param name="prebuilt_db" type="hidden" value="gtdb_genome_reps"/> |
| 274 <param name="prebuilt_date" type="hidden" value="2024-11-09"/> | 403 <param name="prebuilt_date" type="hidden" value="2024-11-09"/> |
| 275 </when> | 404 </when> |
| 276 <when value="eupathdb48_20230407"> | 405 <when value="eupathdb48_20230407"> |
| 278 <param name="prebuilt_date" type="hidden" value="2023-04-07"/> | 407 <param name="prebuilt_date" type="hidden" value="2023-04-07"/> |
| 279 </when> | 408 </when> |
| 280 <when value="eupathdb48_20201113"> | 409 <when value="eupathdb48_20201113"> |
| 281 <param name="prebuilt_db" type="hidden" value="eupathdb48"/> | 410 <param name="prebuilt_db" type="hidden" value="eupathdb48"/> |
| 282 <param name="prebuilt_date" type="hidden" value="2020-11-13"/> | 411 <param name="prebuilt_date" type="hidden" value="2020-11-13"/> |
| 412 </when> | |
| 413 </conditional> | |
| 414 </when> | |
| 415 <when value="amplicon_prebuilt"> | |
| 416 <conditional name="prebuilt"> | |
| 417 <param name="xyz" type="select" multiple="false" label="Select pre-built database to download"> | |
| 418 <option value="16S_Greengenes13.5_20200326">Greengenes 13.5</option> | |
| 419 <option value="16S_RDP11.5_20200326">RDP 11.5</option> | |
| 420 <option value="16S_Silva132_20200326">Silva 132</option> | |
| 421 <option value="16S_Silva138_20200326">Silva 138</option> | |
| 422 </param> | |
| 423 <when value="16S_Greengenes13.5_20200326"> | |
| 424 <param name="prebuilt_db" type="hidden" value="16S_Greengenes13.5"/> | |
| 425 <param name="prebuilt_date" type="hidden" value="20200326"/> | |
| 426 </when> | |
| 427 <when value="16S_RDP11.5_20200326"> | |
| 428 <param name="prebuilt_db" type="hidden" value="16S_RDP11.5"/> | |
| 429 <param name="prebuilt_date" type="hidden" value="20200326"/> | |
| 430 </when> | |
| 431 <when value="16S_Silva132_20200326"> | |
| 432 <param name="prebuilt_db" type="hidden" value="16S_Silva132"/> | |
| 433 <param name="prebuilt_date" type="hidden" value="20200326"/> | |
| 434 </when> | |
| 435 <when value="16S_Silva138_20200326"> | |
| 436 <param name="prebuilt_db" type="hidden" value="16S_Silva138"/> | |
| 437 <param name="prebuilt_date" type="hidden" value="20200326"/> | |
| 283 </when> | 438 </when> |
| 284 </conditional> | 439 </conditional> |
| 285 </when> | 440 </when> |
| 286 <when value="minikraken"> | 441 <when value="minikraken"> |
| 287 <param name="minikraken2_version" type="select" multiple="false" label="Select MiniKraken2 database version to download"> | 442 <param name="minikraken2_version" type="select" multiple="false" label="Select MiniKraken2 database version to download"> |
| 291 </when> | 446 </when> |
| 292 <when value="special"> | 447 <when value="special"> |
| 293 <param name="special_database_type" type="select" multiple="false" label="Select database to build"> | 448 <param name="special_database_type" type="select" multiple="false" label="Select database to build"> |
| 294 <option value="greengenes">Greengenes</option> | 449 <option value="greengenes">Greengenes</option> |
| 295 <option value="silva">Silva</option> | 450 <option value="silva">Silva</option> |
| 296 <option value="rdp">RDP</option> | 451 <!-- <option value="rdp">RDP</option> https://github.com/DerrickWood/kraken2/issues/736 --> |
| 297 </param> | 452 </param> |
| 298 <expand macro="common_params" /> | 453 <expand macro="common_params" /> |
| 299 </when> | 454 </when> |
| 300 <when value="custom"> | 455 <when value="custom"> |
| 301 <param name="custom_fasta" type="data" format="fasta" multiple="False" label="Select history item" /> | 456 <param name="custom_fasta" type="data" format="fasta" multiple="False" label="Select history item" /> |
| 303 <param name="custom_source_info" type="text" label="Database source info" help="Concise description of how this build has been sourced. This description will be appended (in parentheses) to the user-facing name of the build. Example: https://doi.org/10.5281/zenodo.8339822, from v1 assembly_summary.txt sequences" /> | 458 <param name="custom_source_info" type="text" label="Database source info" help="Concise description of how this build has been sourced. This description will be appended (in parentheses) to the user-facing name of the build. Example: https://doi.org/10.5281/zenodo.8339822, from v1 assembly_summary.txt sequences" /> |
| 304 <param name="skip_maps" type="boolean" truevalue="--skip-maps" falsevalue="" label="Skip downloading accession number to taxid maps during taxonomy download." /> | 459 <param name="skip_maps" type="boolean" truevalue="--skip-maps" falsevalue="" label="Skip downloading accession number to taxid maps during taxonomy download." /> |
| 305 <expand macro="common_params" /> | 460 <expand macro="common_params" /> |
| 306 </when> | 461 </when> |
| 307 </conditional> | 462 </conditional> |
| 463 <param name="run_test_command" type="hidden"/> | |
| 308 </inputs> | 464 </inputs> |
| 309 <outputs> | 465 <outputs> |
| 310 <data name="out_file" format="data_manager_json" /> | 466 <data name="out_file" format="data_manager_json" /> |
| 311 </outputs> | 467 </outputs> |
| 312 <tests> | 468 <tests> |
| 469 <!-- standard_local_build --> | |
| 470 | |
| 471 <test expect_num_outputs="1"> | |
| 472 <conditional name="database_type"> | |
| 473 <param name="database_type" value="standard_local_build" /> | |
| 474 <param name="kmer_len" value="35" /> | |
| 475 <param name="minimizer_spaces" value="6"/> | |
| 476 <param name="load_factor" value="0.7" /> | |
| 477 <param name="clean" value="true"/> | |
| 478 </conditional> | |
| 479 <param name="run_test_command" value="false"/> | |
| 480 <output name="out_file"> | |
| 481 <assert_contents> | |
| 482 <has_text text="kraken2_databases"/> | |
| 483 <has_text text="path"/> | |
| 484 <has_text text="Standard (Local Build)"/> | |
| 485 <has_text text="kmer-len=35, minimizer-len=31, minimizer-spaces=6, load-factor=0.7"/> | |
| 486 </assert_contents> | |
| 487 </output> | |
| 488 </test> | |
| 489 | |
| 490 <!-- standard_prebuilt --> | |
| 491 <test> | |
| 492 <conditional name="database_type"> | |
| 493 <param name="database_type" value="standard_prebuilt" /> | |
| 494 <conditional name="prebuilt"> | |
| 495 <param name="prebuilt_date" value="2022-06-07"/> | |
| 496 <param name="prebuilt_db" value="viral"/> | |
| 497 </conditional> | |
| 498 </conditional> | |
| 499 <param name="run_test_command" value="true"/> | |
| 500 <output name="out_file"> | |
| 501 <assert_contents> | |
| 502 <has_text text="kraken2_databases"/> | |
| 503 <has_text text="path"/> | |
| 504 <has_text text="Viral"/> | |
| 505 <has_text text="_standard_prebuilt_viral_2022-06-07"/> | |
| 506 <has_text text="Version"/> | |
| 507 <has_text text="Downloaded"/> | |
| 508 </assert_contents> | |
| 509 </output> | |
| 510 </test> | |
| 511 <!-- test with latest version --> | |
| 512 <test> | |
| 513 <conditional name="database_type"> | |
| 514 <param name="database_type" value="standard_prebuilt" /> | |
| 515 <conditional name="prebuilt"> | |
| 516 <param name="prebuilt_date" value="2024-01-12"/> | |
| 517 <param name="prebuilt_db" value="viral"/> | |
| 518 </conditional> | |
| 519 </conditional> | |
| 520 <param name="run_test_command" value="true"/> | |
| 521 <output name="out_file"> | |
| 522 <assert_contents> | |
| 523 <has_text text="kraken2_databases"/> | |
| 524 <has_text text="path"/> | |
| 525 <has_text text="Viral"/> | |
| 526 <has_text text="_standard_prebuilt_viral_2024-01-12"/> | |
| 527 <has_text text="Version"/> | |
| 528 <has_text text="Downloaded"/> | |
| 529 </assert_contents> | |
| 530 </output> | |
| 531 </test> | |
| 532 <!-- test with first 2024 version --> | |
| 533 <test> | |
| 534 <conditional name="database_type"> | |
| 535 <param name="database_type" value="standard_prebuilt" /> | |
| 536 <conditional name="prebuilt"> | |
| 537 <param name="prebuilt_date" value="2024-06-05"/> | |
| 538 <param name="prebuilt_db" value="viral"/> | |
| 539 </conditional> | |
| 540 </conditional> | |
| 541 <param name="run_test_command" value="true"/> | |
| 542 <output name="out_file"> | |
| 543 <assert_contents> | |
| 544 <has_text text="kraken2_databases"/> | |
| 545 <has_text text="path"/> | |
| 546 <has_text text="Viral"/> | |
| 547 <has_text text="_standard_prebuilt_viral_2024-06-05"/> | |
| 548 <has_text text="Version"/> | |
| 549 <has_text text="Downloaded"/> | |
| 550 </assert_contents> | |
| 551 </output> | |
| 552 </test> | |
| 553 | |
| 554 <!-- minikraken --> | |
| 555 | |
| 556 <test> | |
| 557 <conditional name="database_type"> | |
| 558 <param name="database_type" value="minikraken" /> | |
| 559 <param name="minikraken2_version" value="v1"/> | |
| 560 </conditional> | |
| 561 <param name="run_test_command" value="false"/> | |
| 562 <output name="out_file"> | |
| 563 <assert_contents> | |
| 564 <has_text text="kraken2_databases"/> | |
| 565 <has_text text="path"/> | |
| 566 <has_text text="Minikraken2"/> | |
| 567 <has_text text="v1"/> | |
| 568 <has_text text="Created"/> | |
| 569 </assert_contents> | |
| 570 </output> | |
| 571 </test> | |
| 572 | |
| 573 <!-- special_prebuilt --> | |
| 574 | |
| 575 <test> | |
| 576 <conditional name="database_type"> | |
| 577 <param name="database_type" value="special_prebuilt" /> | |
| 578 <conditional name="prebuilt"> | |
| 579 <param name="xyz" value="eupathdb48_20201113"/> | |
| 580 <param name="prebuilt_date" value="2020-11-13"/> | |
| 581 <param name="prebuilt_db" value="eupathdb48"/> | |
| 582 </conditional> | |
| 583 </conditional> | |
| 584 <param name="run_test_command" value="false"/> | |
| 585 <output name="out_file"> | |
| 586 <assert_contents> | |
| 587 <has_text text="kraken2_databases"/> | |
| 588 <has_text text="path"/> | |
| 589 <has_text text="Prebuilt Refseq indexes: EuPathDB-46"/> | |
| 590 <has_text text="standard_prebuilt_eupathdb48_2020-11-13"/> | |
| 591 <has_text text="Prebuilt Refseq indexes"/> | |
| 592 <has_text text="Downloaded"/> | |
| 593 </assert_contents> | |
| 594 </output> | |
| 595 </test> | |
| 596 | |
| 597 <!-- amplicon_prebuilt --> | |
| 598 | |
| 599 <test> | |
| 600 <conditional name="database_type"> | |
| 601 <param name="database_type" value="amplicon_prebuilt" /> | |
| 602 <conditional name="prebuilt"> | |
| 603 <param name="xyz" value="16S_Greengenes13.5_20200326"/> | |
| 604 <param name="prebuilt_date" value="20200326"/> | |
| 605 <param name="prebuilt_db" value="16S_Greengenes13.5"/> | |
| 606 </conditional> | |
| 607 </conditional> | |
| 608 <param name="run_test_command" value="false"/> | |
| 609 <output name="out_file"> | |
| 610 <assert_contents> | |
| 611 <has_text text="kraken2_databases"/> | |
| 612 <has_text text="path"/> | |
| 613 <has_text text="16S_Greengenes13.5"/> | |
| 614 </assert_contents> | |
| 615 </output> | |
| 616 </test> | |
| 617 | |
| 618 <!-- special --> | |
| 619 | |
| 620 <test expect_num_outputs="1"> | |
| 621 <conditional name="database_type"> | |
| 622 <param name="database_type" value="special" /> | |
| 623 <param name="special_database_type" value="greengenes" /> | |
| 624 <param name="kmer_len" value="35" /> | |
| 625 <param name="minimizer_spaces" value="6"/> | |
| 626 <param name="load_factor" value="0.7" /> | |
| 627 <param name="clean" value="true"/> | |
| 628 </conditional> | |
| 629 <param name="run_test_command" value="true"/> | |
| 630 <output name="out_file"> | |
| 631 <assert_contents> | |
| 632 <has_text text="kraken2_databases"/> | |
| 633 <has_text text="path"/> | |
| 634 <has_text text="Greengenes"/> | |
| 635 <has_text text="kmer-len=35, minimizer-len=31, minimizer-spaces=6, load-factor=0.7"/> | |
| 636 </assert_contents> | |
| 637 </output> | |
| 638 </test> | |
| 639 <test expect_num_outputs="1"> | |
| 640 <conditional name="database_type"> | |
| 641 <param name="database_type" value="special" /> | |
| 642 <param name="special_database_type" value="silva" /> | |
| 643 <param name="kmer_len" value="35" /> | |
| 644 <param name="minimizer_spaces" value="6"/> | |
| 645 <param name="load_factor" value="0.7" /> | |
| 646 <param name="clean" value="true"/> | |
| 647 </conditional> | |
| 648 <param name="run_test_command" value="true"/> | |
| 649 <output name="out_file"> | |
| 650 <assert_contents> | |
| 651 <has_text text="kraken2_databases"/> | |
| 652 <has_text text="path"/> | |
| 653 <has_text text="Silva"/> | |
| 654 <has_text text="kmer-len=35, minimizer-len=31, minimizer-spaces=6, load-factor=0.7"/> | |
| 655 </assert_contents> | |
| 656 </output> | |
| 657 </test> | |
| 658 | |
| 659 <!-- custom --> | |
| 660 | |
| 313 <test expect_num_outputs="1"> | 661 <test expect_num_outputs="1"> |
| 314 <conditional name="database_type"> | 662 <conditional name="database_type"> |
| 315 <param name="database_type" value="custom" /> | 663 <param name="database_type" value="custom" /> |
| 316 <param name="custom_fasta" value="adapter.fa" /> | 664 <param name="custom_fasta" value="adapter.fa" /> |
| 317 <param name="custom_database_name" value="custom_database" /> | 665 <param name="custom_database_name" value="custom_database" /> |
| 320 <param name="kmer_len" value="35" /> | 668 <param name="kmer_len" value="35" /> |
| 321 <param name="minimizer_spaces" value="6"/> | 669 <param name="minimizer_spaces" value="6"/> |
| 322 <param name="load_factor" value="0.7" /> | 670 <param name="load_factor" value="0.7" /> |
| 323 <param name="clean" value="true"/> | 671 <param name="clean" value="true"/> |
| 324 </conditional> | 672 </conditional> |
| 673 <param name="run_test_command" value="true"/> | |
| 325 <output name="out_file"> | 674 <output name="out_file"> |
| 326 <assert_contents> | 675 <assert_contents> |
| 327 <has_text text="kraken2_databases"/> | 676 <has_text text="kraken2_databases"/> |
| 328 <has_text text="path"/> | 677 <has_text text="path"/> |
| 329 <has_text text="custom_database (from adapter.fa test data, kmer-len=35, minimizer-len=31, minimizer-spaces=6, load-factor=0.7)"/> | 678 <has_text text="custom_database (from adapter.fa test data, kmer-len=35, minimizer-len=31, minimizer-spaces=6, load-factor=0.7)"/> |
| 330 </assert_contents> | 679 </assert_contents> |
| 331 </output> | 680 </output> |
| 332 </test> | 681 </test> |
| 333 <test> | 682 |
| 334 <conditional name="database_type"> | 683 |
| 335 <param name="database_type" value="standard_prebuilt" /> | |
| 336 <conditional name="prebuild"> | |
| 337 <param name="prebuilt_date" value="2022-06-07"/> | |
| 338 <param name="prebuilt_db" value="viral"/> | |
| 339 </conditional> | |
| 340 </conditional> | |
| 341 <output name="out_file"> | |
| 342 <assert_contents> | |
| 343 <has_text text="kraken2_databases"/> | |
| 344 <has_text text="path"/> | |
| 345 <has_text text="Viral"/> | |
| 346 <has_text text="_standard_prebuilt_viral_2022-06-07"/> | |
| 347 <has_text text="Version"/> | |
| 348 <has_text text="Downloaded"/> | |
| 349 </assert_contents> | |
| 350 </output> | |
| 351 </test> | |
| 352 <!-- test with latest version --> | |
| 353 <test> | |
| 354 <conditional name="database_type"> | |
| 355 <param name="database_type" value="standard_prebuilt" /> | |
| 356 <conditional name="prebuild"> | |
| 357 <param name="prebuilt_date" value="2024-01-12"/> | |
| 358 <param name="prebuilt_db" value="viral"/> | |
| 359 </conditional> | |
| 360 </conditional> | |
| 361 <output name="out_file"> | |
| 362 <assert_contents> | |
| 363 <has_text text="kraken2_databases"/> | |
| 364 <has_text text="path"/> | |
| 365 <has_text text="Viral"/> | |
| 366 <has_text text="_standard_prebuilt_viral_2024-01-12"/> | |
| 367 <has_text text="Version"/> | |
| 368 <has_text text="Downloaded"/> | |
| 369 </assert_contents> | |
| 370 </output> | |
| 371 </test> | |
| 372 <!-- test with first 2024 version --> | |
| 373 <test> | |
| 374 <conditional name="database_type"> | |
| 375 <param name="database_type" value="standard_prebuilt" /> | |
| 376 <conditional name="prebuild"> | |
| 377 <param name="prebuilt_date" value="2024-06-05"/> | |
| 378 <param name="prebuilt_db" value="viral"/> | |
| 379 </conditional> | |
| 380 </conditional> | |
| 381 <output name="out_file"> | |
| 382 <assert_contents> | |
| 383 <has_text text="kraken2_databases"/> | |
| 384 <has_text text="path"/> | |
| 385 <has_text text="Viral"/> | |
| 386 <has_text text="_standard_prebuilt_viral_2024-06-05"/> | |
| 387 <has_text text="Version"/> | |
| 388 <has_text text="Downloaded"/> | |
| 389 </assert_contents> | |
| 390 </output> | |
| 391 </test> | |
| 392 </tests> | 684 </tests> |
| 393 <help><![CDATA[ | 685 <help><![CDATA[ |
| 394 Build Kraken2 databases or download `prebuilt Kraken2 RefSeq indexes <https://benlangmead.github.io/aws-indexes/k2>`__ | 686 Build Kraken2 databases or download `prebuilt Kraken2 RefSeq indexes <https://benlangmead.github.io/aws-indexes/k2>`__ |
| 395 ]]></help> | 687 ]]></help> |
| 396 <citations> | 688 <citations> |
