| Previous changeset 5:2f27f3b86827 (2021-11-08) Next changeset 7:ed1518ce2237 (2023-04-16) |
|
Commit message:
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_kraken2_database/ commit 5d74f13e59c4a2862c108ac1a08c067b0cfb2d97 |
|
modified:
data_manager/kraken2_build_database.py data_manager/kraken2_build_database.xml |
| b |
| diff -r 2f27f3b86827 -r 9002633b4737 data_manager/kraken2_build_database.py --- a/data_manager/kraken2_build_database.py Mon Nov 08 15:40:34 2021 +0000 +++ b/data_manager/kraken2_build_database.py Fri Jun 24 12:44:33 2022 +0000 |
| [ |
| @@ -54,9 +54,17 @@ class StandardPrebuiltSizes(Enum): - full = 'full' - gb_16 = '16' - gb_8 = '8' + viral = "viral" + minusb = "minusb" + standard = "standard" + standard_08gb = "standard_08gb" + standard_16gb = "standard_16gb" + pluspf = "pluspf" + pluspf_08gb = "pluspf_08gb" + pluspf_16gb = "pluspf_16gb" + pluspfp = "pluspfp" + pluspfp_08gb = "pluspfp_08gb" + pluspfp_16gb = "pluspfp_16gb" def __str__(self): return self.value @@ -122,36 +130,47 @@ return data_table_entry -def kraken2_build_standard_prebuilt(standard_prebuilt_size, prebuilt_date, target_directory, data_table_name=DATA_TABLE_NAME): +def kraken2_build_standard_prebuilt(prebuilt_db, prebuilt_date, target_directory, data_table_name=DATA_TABLE_NAME): now = datetime.datetime.utcnow().strftime("%Y-%m-%dT%H%M%SZ") + prebuild_name = { + 'viral': "Viral", + 'minusb': "MinusB (archaea, viral, plasmid, human, UniVec_Core)", + 'standard': "Standard-Full (archaea, bacteria, viral, plasmid, human,UniVec_Core)", + 'standard_08gb': "Standard-8 (Standard with DB capped at 8 GB)", + 'standard_16gb': "Standard-16 (Standard with DB capped at 16 GB)", + 'pluspf': "PlusPF (Standard plus protozoa and fungi)", + 'pluspf_08gb': "PlusPF-8 (PlusPF with DB capped at 8 GB)", + 'pluspf_16gb': "PlusPF-16 (PlusPF with DB capped at 16 GB)", + 'pluspfp': "PlusPFP (Standard plus protozoa, fungi and plant)", + 'pluspfp_08gb': "PlusPFP-8 (PlusPFP with DB capped at 8 GB)", + 'pluspfp_16gb': "PlusPFP-16 (PlusPFP with DB capped at 16 GB)" + } + database_value = "_".join([ now, "standard_prebuilt", - standard_prebuilt_size + prebuilt_db, + prebuilt_date ]) database_name = " ".join([ - "Standard (Prebuilt)", - standard_prebuilt_size, - "(Downloaded:", + "Prebuilt Refseq indexes: ", + prebuild_name[prebuilt_db], + "(Version: ", + prebuilt_date, + "- Downloaded:", now + ")" ]) database_path = database_value - size_to_url_str = { - 'full': '', - '16': '_16gb', - '8': '_8gb', - } # we may need to let the user choose the date when new DBs are posted. date_url_str = prebuilt_date.replace('-', '') - standard_prebuilt_size_url = size_to_url_str[standard_prebuilt_size] # download the pre-built database try: - download_url = 'https://genome-idx.s3.amazonaws.com/kraken/k2_standard%s_%s.tar.gz' % (standard_prebuilt_size_url, date_url_str) + download_url = 'https://genome-idx.s3.amazonaws.com/kraken/k2_%s_%s.tar.gz' % (prebuilt_db, date_url_str) src = urlopen(download_url) except URLError as e: print('url: ' + download_url, file=sys.stderr) @@ -369,7 +388,7 @@ parser.add_argument('--threads', dest='threads', default=1, help='threads') parser.add_argument('--database-type', dest='database_type', type=KrakenDatabaseTypes, choices=list(KrakenDatabaseTypes), required=True, help='type of kraken database to build') parser.add_argument('--minikraken2-version', dest='minikraken2_version', type=Minikraken2Versions, choices=list(Minikraken2Versions), help='MiniKraken2 version (only applies to --database-type minikraken)') - parser.add_argument('--standard-prebuilt-size', dest='standard_prebuilt_size', type=StandardPrebuiltSizes, choices=list(StandardPrebuiltSizes), help='Size of standard prebuilt database to download (only applies to --database-type standard_prebuilt. Options are: "8", "16", "full".)') + parser.add_argument('--prebuilt-db', dest='prebuilt_db', type=StandardPrebuiltSizes, choices=list(StandardPrebuiltSizes), help='Prebuilt database to download. Only applies to --database-type standard_prebuilt.') parser.add_argument('--prebuilt-date', dest='prebuilt_date', help='Database build date (YYYY-MM-DD). Only applies to --database-type standard_prebuilt.') parser.add_argument('--special-database-type', dest='special_database_type', type=SpecialDatabaseTypes, choices=list(SpecialDatabaseTypes), help='type of special database to build (only applies to --database-type special)') parser.add_argument('--custom-fasta', dest='custom_fasta', help='fasta file for custom database (only applies to --database-type custom)') @@ -408,7 +427,7 @@ ) elif str(args.database_type) == 'standard_prebuilt': data_manager_output = kraken2_build_standard_prebuilt( - str(args.standard_prebuilt_size), + str(args.prebuilt_db), str(args.prebuilt_date), target_directory ) |
| b |
| diff -r 2f27f3b86827 -r 9002633b4737 data_manager/kraken2_build_database.xml --- a/data_manager/kraken2_build_database.xml Mon Nov 08 15:40:34 2021 +0000 +++ b/data_manager/kraken2_build_database.xml Fri Jun 24 12:44:33 2022 +0000 |
| [ |
| b'@@ -1,7 +1,9 @@\n <?xml version="1.0"?>\n-<tool id="kraken2_build_database" name="Kraken2" tool_type="manage_data" version="@TOOL_VERSION@" profile="18.09">\n+<tool id="kraken2_build_database" name="Kraken2" tool_type="manage_data" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">\n <macros>\n- <token name="@TOOL_VERSION@">2.1.1</token>\n+ <token name="@TOOL_VERSION@">2.1.2</token>\n+ <token name="@VERSION_SUFFIX@">0</token>\n+ <token name="@PROFILE@">21.01</token>\n <xml name="common_params">\n <param name="kmer_len" type="integer" value="35" label="K-mer length in BP" />\n <param name="minimizer_len" type="integer" value="31" label="Minimizer length" />\n@@ -9,6 +11,39 @@\n <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" />\n <param name="clean" type="boolean" truevalue="--clean" falsevalue="" checked="true" label="Clean up extra files" />\n </xml>\n+ <xml name="viral">\n+ <option value="viral">Viral (viral; ~50 GB)</option>\n+ </xml>\n+ <xml name="minusb">\n+ <option value="minusb">MinusB (archaea, viral, plasmid, human, UniVec_Core; ~8 GB)</option>\n+ </xml>\n+ <xml name="standard">\n+ <option value="standard">Standard-Full (archaea, bacteria, viral, plasmid, human, UniVec_Core; ~50 GB)</option>\n+ </xml>\n+ <xml name="standard_08gb">\n+ <option value="standard_08gb">Standard-8 (Standard with DB capped at 8 GB)</option>\n+ </xml>\n+ <xml name="standard_16gb">\n+ <option value="standard_16gb">Standard-16 (Standard with DB capped at 16 GB)</option>\n+ </xml>\n+ <xml name="pluspf">\n+ <option value="pluspf">PlusPF (Standard plus protozoa and fungi; ~50 GB)</option>\n+ </xml>\n+ <xml name="pluspf_08gb">\n+ <option value="pluspf_08gb">PlusPF-8 (PlusPF with DB capped at 8 GB; ~7.5 GB)</option>\n+ </xml>\n+ <xml name="pluspf_16gb">\n+ <option value="pluspf_16gb">PlusPF-16 (PlusPF with DB capped at 16 GB; ~15 GB)</option>\n+ </xml>\n+ <xml name="pluspfp">\n+ <option value="pluspfp">PlusPFP (Standard plus protozoa, fungi and plant; ~129 GB)</option>\n+ </xml>\n+ <xml name="pluspfp_08gb">\n+ <option value="pluspfp_08gb">PlusPFP-8 (PlusPFP with DB capped at 8 GB; ~7.5 GB)</option>\n+ </xml>\n+ <xml name="pluspfp_16gb">\n+ <option value="pluspfp_16gb">PlusPFP-16 (PlusPFP with DB capped at 16 GB; ~15 GB)</option>\n+ </xml>\n </macros>\n <description>database builder</description>\n <requirements>\n@@ -16,53 +51,48 @@\n <requirement type="package" version="3.7">python</requirement>\n </requirements>\n <version_command>kraken2 -version | head -n 1 | awk \'{print $NF}\'</version_command>\n- <command detect_errors="exit_code">\n- <![CDATA[\n- python \'$__tool_directory__/kraken2_build_database.py\'\n- \'${out_file}\'\n- --database-type ${database_type.database_type}\n- #if $database_type.database_type == "minikraken"\n- --minikraken2-version ${database_type.minikraken2_version}\n- #else if $database_type.database_type == "standard_prebuilt"\n- --standard-prebuilt-size ${database_type.standard_prebuilt_size}\n- #if $database_type.prebuilt_date_custom != ""\n- --prebuilt-date ${database_type.prebuilt_date_custom}\n- #else\n- --prebuilt-date ${database_type.prebuilt_date}\n- #end if\n- #else if $database_type.database_type == "standard_local_build"\n- --threads \\${GALAXY_SLOTS:-1}\n- --kmer-len ${database_type.kmer_len}\n- --minimizer-len ${database_type.minimizer_len}\n- --minimizer-spaces ${database_type.minimizer_spaces}\n- --lo'..b'o="pluspf"/>\n+ <expand macro="pluspf_08gb"/>\n+ <expand macro="pluspf_16gb"/>\n+ <expand macro="pluspfp_08gb"/>\n+ <expand macro="pluspfp_16gb"/>\n+ </param>\n+ </when>\n+ </conditional>\n </when>\n <when value="minikraken">\n <param name="minikraken2_version" type="select" multiple="false" label="Select MiniKraken2 database version to download">\n@@ -97,9 +186,9 @@\n <expand macro="common_params" />\n </when>\n <when value="custom">\n- <param name="custom_fasta" type="data" format="fasta" multiple="False" optional="true" label="Select history item" />\n+ <param name="custom_fasta" type="data" format="fasta" multiple="False" label="Select history item" />\n <param name="custom_database_name" type="text" label="Name for this database" />\n- <param name="skip_maps" type="boolean" truevalue="--skip-maps" falsevalue="" label="Skip downloading accession number to taxid maps during taxonomy download." /> \n+ <param name="skip_maps" type="boolean" truevalue="--skip-maps" falsevalue="" label="Skip downloading accession number to taxid maps during taxonomy download." />\n <expand macro="common_params" />\n </when>\n </conditional>\n@@ -108,16 +197,48 @@\n <data name="out_file" format="data_manager_json" />\n </outputs>\n <tests>\n+ <test expect_num_outputs="1">\n+ <conditional name="database_type">\n+ <param name="database_type" value="custom" />\n+ <param name="custom_fasta" value="adapter.fa" />\n+ <param name="custom_database_name" value="custom_database" />\n+ <param name="skip_maps" value="true" />\n+ <param name="kmer_len" value="35" />\n+ <param name="minimizer_spaces" value="6"/>\n+ <param name="load_factor" value="0.7" />\n+ <param name="clean" truevalue="--clean"/>\n+ </conditional>\n+ <output name="out_file">\n+ <assert_contents>\n+ <has_text text="kraken2_databases"/>\n+ <has_text text="path"/>\n+ <has_text text="custom_database"/>\n+ </assert_contents>\n+ </output>\n+ </test>\n <test>\n- <param name="database_type" value="custom" />\n- <param name="custom_fasta" value="adapter.fa" />\n- <param name="custom_database_name" value="database" />\n- <param name="skip_maps" value="true" />\n- <output name="out_file" value="kraken2_custom_data_manager.json" />\n+ <conditional name="database_type">\n+ <param name="database_type" value="standard_prebuilt" />\n+ <conditional name="prebuild">\n+ <param name="prebuilt_date" value="2022-06-07"/>\n+ <param name="prebuilt_db" value="standard_08gb"/>\n+ </conditional>\n+ </conditional>\n+ <output name="out_file">\n+ <assert_contents>\n+ <has_text text="kraken2_databases"/>\n+ <has_text text="path"/>\n+ <has_text text="Standard-8"/>\n+ <has_text text="_standard_prebuilt_standard_08gb_2022-06-07"/>\n+ <has_text text="Version"/>\n+ <has_text text="Downloaded"/>\n+ </assert_contents>\n+ </output>\n </test>\n </tests>\n- <help>\n- </help>\n+ <help><![CDATA[\n+Build Kraken2 databases or download `prebuilt Kraken2 RefSeq indexes <https://benlangmead.github.io/aws-indexes/k2>`__\n+ ]]></help>\n <citations>\n <citation type="doi">10.1186/gb-2014-15-3-r46</citation>\n </citations>\n' |