Mercurial > repos > dvanzessen > vep_emc
changeset 6:3333d43a5858 draft
Uploaded
| author | dvanzessen |
|---|---|
| date | Tue, 16 Jul 2019 04:57:05 -0400 |
| parents | 166e1c2b8c35 |
| children | 01748ca044e6 |
| files | data_manager/fetch_vep_cache_data.py data_manager/fetch_vep_cache_data.xml data_manager_conf.xml tool-data/vep_cache_data.loc.sample tool_data_table_conf.xml.sample vep.xml |
| diffstat | 6 files changed, 9 insertions(+), 305 deletions(-) [+] |
line wrap: on
line diff
--- a/data_manager/fetch_vep_cache_data.py Mon Jul 15 09:37:13 2019 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,70 +0,0 @@ -import argparse -import os -import json -import re -import pprint -import subprocess -import sys - -if __name__ == "__main__": - parser = argparse.ArgumentParser(description="Use VEP INSTALL.pl to download/process the cache for an assembly") - parser.add_argument("--output-file") - parser.add_argument("--output-dir") - parser.add_argument("--species") - parser.add_argument("--species-type", choices=["ensembl", "refseq", "merged"], default="ensembl") - args = parser.parse_args() - - output_file = args.output_file - output_dir = args.output_dir - - species = args.species - assembly = "" - - if species.startswith("homo_sapiens"): - if species.endswith("37"): - assembly = " --ASSEMBLY GRCh37" - elif species.endswith("38"): - assembly = " --ASSEMBLY GRCh38" - else: - print("Unknown human assembly") - sys.exit(1) - species = "homo_sapiens" - - species_type = args.species_type - if species_type in ["refseq", "merged"]: - species = "{0}_{1}".format(species, species_type) - - with open(output_file) as output_file_handle: - params = json.loads(output_file_handle.read()) - - print(output_file) - print(output_dir) - print(species) - print(species_type) - pprint.pprint(params) - - if not os.path.exists(output_dir): - os.makedirs(output_dir) - - vep_install_cmd = "vep_install --NO_HTSLIB -a alcf --CACHEDIR {0} --SPECIES {1}{2}".format(output_dir, species, assembly) - - print("Running INSTALL.pl") - print(vep_install_cmd) - exit_code = subprocess.call(vep_install_cmd, cwd=output_dir, shell=True) - - print(exit_code) - - output_dict = dict( - data_tables=dict( - vep_cache_data=[{ - "value": species, - "path": output_dir, - "dbkey": args.species, - "type": species_type, - "name": "{0} ({1})".format(args.species, species_type) - }] - ) - ) - with open(output_file, 'w') as output_file_handle: - output_file_handle.write(json.dumps(output_dict)) - sys.exit(exit_code) \ No newline at end of file
--- a/data_manager/fetch_vep_cache_data.xml Mon Jul 15 09:37:13 2019 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,185 +0,0 @@ -<tool id="fetch_vep_cache_data" name="VEP Cache" version="0.0.1" tool_type="manage_data"> - <requirements> - <requirement type="package" version="94.5">ensembl-vep</requirement> - <requirement type="package" version="1.642">perl-dbi</requirement> - <requirement type="package" version="1.64">perl-archive-zip</requirement> - <requirement type="package" version="4.046">perl-dbd-mysql</requirement> - </requirements> - <description>fetching</description> - - <command detect_errors="exit_code"> -python $__tool_directory__/fetch_vep_cache_data.py ---output-file ${ out_file } ---output-dir ${ out_file.extra_files_path } ---species ${ species } ---species-type ${ species_type } - </command> - <inputs> - <param name="species" type="select" label="Species"> - <option value="homo_sapiens37">homo_sapiens (GRCh37)</option> - <option value="homo_sapiens38">homo_sapiens (GRCh38)</option> - <option value="orycteropus_afer_afer">orycteropus_afer_afer</option> - <option value="mus_spretus">mus_spretus</option> - <option value="vicugna_pacos">vicugna_pacos</option> - <option value="poecilia_formosa">poecilia_formosa</option> - <option value="colobus_angolensis_palliatus">colobus_angolensis_palliatus</option> - <option value="anolis_carolinensis">anolis_carolinensis</option> - <option value="dasypus_novemcinctus">dasypus_novemcinctus</option> - <option value="scleropages_formosus">scleropages_formosus</option> - <option value="labrus_bergylta">labrus_bergylta</option> - <option value="stegastes_partitus">stegastes_partitus</option> - <option value="periophthalmus_magnuspinnatus">periophthalmus_magnuspinnatus</option> - <option value="rhinopithecus_bieti">rhinopithecus_bieti</option> - <option value="saimiri_boliviensis_boliviensis">saimiri_boliviensis_boliviensis</option> - <option value="pan_paniscus">pan_paniscus</option> - <option value="cavia_aperea">cavia_aperea</option> - <option value="melopsittacus_undulatus">melopsittacus_undulatus</option> - <option value="haplochromis_burtoni">haplochromis_burtoni</option> - <option value="otolemur_garnettii">otolemur_garnettii</option> - <option value="ciona_intestinalis">ciona_intestinalis</option> - <option value="ciona_savignyi">ciona_savignyi</option> - <option value="caenorhabditis_elegans">caenorhabditis_elegans</option> - <option value="cebus_capucinus_imitator">cebus_capucinus_imitator</option> - <option value="felis_catus">felis_catus</option> - <option value="astyanax_mexicanus">astyanax_mexicanus</option> - <option value="ictalurus_punctatus">ictalurus_punctatus</option> - <option value="gallus_gallus">gallus_gallus</option> - <option value="pan_troglodytes">pan_troglodytes</option> - <option value="cricetulus_griseus">cricetulus_griseus</option> - <option value="pelodiscus_sinensis">pelodiscus_sinensis</option> - <option value="anabas_testudineus">anabas_testudineus</option> - <option value="amphiprion_ocellaris">amphiprion_ocellaris</option> - <option value="gadus_morhua">gadus_morhua</option> - <option value="latimeria_chalumnae">latimeria_chalumnae</option> - <option value="propithecus_coquereli">propithecus_coquereli</option> - <option value="bos_taurus">bos_taurus</option> - <option value="macaca_fascicularis">macaca_fascicularis</option> - <option value="fukomys_damarensis">fukomys_damarensis</option> - <option value="octodon_degus">octodon_degus</option> - <option value="canis_lupus_familiaris">canis_lupus_familiaris</option> - <option value="tursiops_truncatus">tursiops_truncatus</option> - <option value="mandrillus_leucophaeus">mandrillus_leucophaeus</option> - <option value="anas_platyrhynchos">anas_platyrhynchos</option> - <option value="astatotilapia_calliptera">astatotilapia_calliptera</option> - <option value="loxodonta_africana">loxodonta_africana</option> - <option value="mustela_putorius_furo">mustela_putorius_furo</option> - <option value="ficedula_albicollis">ficedula_albicollis</option> - <option value="drosophila_melanogaster">drosophila_melanogaster</option> - <option value="takifugu_rubripes">takifugu_rubripes</option> - <option value="nomascus_leucogenys">nomascus_leucogenys</option> - <option value="capra_hircus">capra_hircus</option> - <option value="mesocricetus_auratus">mesocricetus_auratus</option> - <option value="rhinopithecus_roxellana">rhinopithecus_roxellana</option> - <option value="gorilla_gorilla_gorilla">gorilla_gorilla_gorilla</option> - <option value="seriola_dumerili">seriola_dumerili</option> - <option value="cavia_porcellus">cavia_porcellus</option> - <option value="poecilia_reticulata">poecilia_reticulata</option> - <option value="eptatretus_burgeri">eptatretus_burgeri</option> - <option value="papio_hamadryas">papio_hamadryas</option> - <option value="erinaceus_europaeus">erinaceus_europaeus</option> - <option value="equus_caballus">equus_caballus</option> - <option value="procavia_capensis">procavia_capensis</option> - <option value="oryzias_melastigma">oryzias_melastigma</option> - <option value="oryzias_latipes">oryzias_latipes</option> - <option value="dipodomys_ordii">dipodomys_ordii</option> - <option value="petromyzon_marinus">petromyzon_marinus</option> - <option value="panthera_pardus">panthera_pardus</option> - <option value="jaculus_jaculus">jaculus_jaculus</option> - <option value="echinops_telfairi">echinops_telfairi</option> - <option value="chinchilla_lanigera">chinchilla_lanigera</option> - <option value="neolamprologus_brichardi">neolamprologus_brichardi</option> - <option value="aotus_nancymaae">aotus_nancymaae</option> - <option value="macaca_mulatta">macaca_mulatta</option> - <option value="kryptolebias_marmoratus">kryptolebias_marmoratus</option> - <option value="callithrix_jacchus">callithrix_jacchus</option> - <option value="pteropus_vampyrus">pteropus_vampyrus</option> - <option value="myotis_lucifugus">myotis_lucifugus</option> - <option value="amphilophus_citrinellus">amphilophus_citrinellus</option> - <option value="xiphophorus_couchianus">xiphophorus_couchianus</option> - <option value="mus_musculus">mus_musculus</option> - <option value="mus_musculus_castaneus">mus_musculus_castaneus</option> - <option value="microcebus_murinus">microcebus_murinus</option> - <option value="mus_musculus_musculus">mus_musculus_musculus</option> - <option value="mus_musculus_domesticus">mus_musculus_domesticus</option> - <option value="fundulus_heteroclitus">fundulus_heteroclitus</option> - <option value="heterocephalus_glaber">heterocephalus_glaber</option> - <option value="peromyscus_maniculatus_bairdii">peromyscus_maniculatus_bairdii</option> - <option value="esox_lucius">esox_lucius</option> - <option value="mola_mola">mola_mola</option> - <option value="papio_anubis">papio_anubis</option> - <option value="monodelphis_domestica">monodelphis_domestica</option> - <option value="amphiprion_percula">amphiprion_percula</option> - <option value="pongo_abelii">pongo_abelii</option> - <option value="chrysemys_picta_bellii">chrysemys_picta_bellii</option> - <option value="ailuropoda_melanoleuca">ailuropoda_melanoleuca</option> - <option value="paramormyrops_kingsleyae">paramormyrops_kingsleyae</option> - <option value="sus_scrofa">sus_scrofa</option> - <option value="sus_scrofa_map">sus_scrofa_map</option> - <option value="macaca_nemestrina">macaca_nemestrina</option> - <option value="ochotona_princeps">ochotona_princeps</option> - <option value="xiphophorus_maculatus">xiphophorus_maculatus</option> - <option value="ornithorhynchus_anatinus">ornithorhynchus_anatinus</option> - <option value="microtus_ochrogaster">microtus_ochrogaster</option> - <option value="pundamilia_nyererei">pundamilia_nyererei</option> - <option value="oryctolagus_cuniculus">oryctolagus_cuniculus</option> - <option value="rattus_norvegicus">rattus_norvegicus</option> - <option value="pygocentrus_nattereri">pygocentrus_nattereri</option> - <option value="ceratotherium_simum_simum">ceratotherium_simum_simum</option> - <option value="mus_caroli">mus_caroli</option> - <option value="saccharomyces_cerevisiae">saccharomyces_cerevisiae</option> - <option value="poecilia_latipinna">poecilia_latipinna</option> - <option value="ovis_aries">ovis_aries</option> - <option value="cyprinodon_variegatus">cyprinodon_variegatus</option> - <option value="poecilia_mexicana">poecilia_mexicana</option> - <option value="sorex_araneus">sorex_araneus</option> - <option value="mus_pahari">mus_pahari</option> - <option value="choloepus_hoffmanni">choloepus_hoffmanni</option> - <option value="cercocebus_atys">cercocebus_atys</option> - <option value="physter_macrocephalus">physter_macrocephalus</option> - <option value="acanthochromis_polyacanthus">acanthochromis_polyacanthus</option> - <option value="lepisosteus_oculatus">lepisosteus_oculatus</option> - <option value="ictidomys_tridecemlineatus">ictidomys_tridecemlineatus</option> - <option value="gasterosteus_aculeatus">gasterosteus_aculeatus</option> - <option value="monopterus_albus">monopterus_albus</option> - <option value="carlito_syrichta">carlito_syrichta</option> - <option value="sarcophilus_harrisii">sarcophilus_harrisii</option> - <option value="tetraodon_nigroviridis">tetraodon_nigroviridis</option> - <option value="panthera_tigris_altaica">panthera_tigris_altaica</option> - <option value="hippocampus_comes">hippocampus_comes</option> - <option value="oreochromis_niloticus">oreochromis_niloticus</option> - <option value="cynoglossus_semilaevis">cynoglossus_semilaevis</option> - <option value="tupaia_belangeri">tupaia_belangeri</option> - <option value="scophthalmus_maximus">scophthalmus_maximus</option> - <option value="meleagris_gallopavo">meleagris_gallopavo</option> - <option value="nannospalax_galili">nannospalax_galili</option> - <option value="chlorocebus_sabaeus">chlorocebus_sabaeus</option> - <option value="notamacropus_eugenii">notamacropus_eugenii</option> - <option value="gambusia_affinis">gambusia_affinis</option> - <option value="xenopus_tropicalis">xenopus_tropicalis</option> - <option value="seriola_lalandi_dorsalis">seriola_lalandi_dorsalis</option> - <option value="taeniopygia_guttata">taeniopygia_guttata</option> - <option value="maylandia_zebra">maylandia_zebra</option> - <option value="danio_rerio">danio_rerio</option> - <option value="mastacembelus_armatus">mastacembelus_armatus</option> - <option value=""></option> - </param> - <param name="species_type" type="select" label="Get the Ensembl, Refseq or merged Ensembl/Refseq"> - <option value="ensembl" selected="true">Ensembl</option> - <option value="refseq">Refseq</option> - <option value="merged">Merged</option> - </param> - </inputs> - <outputs> - <data name="out_file" format="data_manager_json"/> - </outputs> - <!-- - <tests> - <test> - DON'T FORGET TO DEFINE SOME TOOL TESTS - </test> - </tests> - --> - <help> - There is no help - </help> -</tool>
--- a/data_manager_conf.xml Mon Jul 15 09:37:13 2019 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,20 +0,0 @@ -<?xml version="1.0"?> -<data_managers> - <data_manager tool_file="data_manager/fetch_vep_cache_data.xml" id="fetch_vep_cache_data"> - <data_table name="vep_cache_data"> - <output> - <column name="value" /> - <column name="dbkey" /> - <column name="type" /> - <column name="name" /> - <column name="path" output_ref="out_file" > - <move type="directory"> - <source>${path}</source> - <target base="${GALAXY_DATA_MANAGER_DATA_PATH}">${dbkey}/seq/${path}</target> - </move> - <value_translation>${GALAXY_DATA_MANAGER_DATA_PATH}/${dbkey}/seq/${path}</value_translation> - </column> - </output> - </data_table> - </data_manager> -</data_managers>
--- a/tool-data/vep_cache_data.loc.sample Mon Jul 15 09:37:13 2019 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,7 +0,0 @@ -# VEP data cache location -# <value> <dbkey> <type> <name> <path> -# <value> unique key of an entry (species_<type>) -# <dbkey> galaxy ref key -# <type> cache type, ensembl, refseq or merged -# <name> display name of the entry -# <path> path to the directory of the cache dir \ No newline at end of file
--- a/tool_data_table_conf.xml.sample Mon Jul 15 09:37:13 2019 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,7 +0,0 @@ -<?xml version="1.0"?> -<tables> - <table name="vep_cache_data" comment_char="#" allow_duplicate_entries="False"> - <columns>value, dbkey, type, name, path</columns> - <file path="tool-data/vep_cache_data.loc" /> - </table> -</tables>
--- a/vep.xml Mon Jul 15 09:37:13 2019 -0400 +++ b/vep.xml Tue Jul 16 04:57:05 2019 -0400 @@ -1,19 +1,8 @@ <tool id="vep" name="Vep" version="0.1.0"> <requirements> <requirement type="package" version="94.5">ensembl-vep</requirement> - <requirement type="package" version="1.642">perl-dbi</requirement> - <requirement type="package" version="1.64">perl-archive-zip</requirement> - <requirement type="package" version="4.046">perl-dbd-mysql</requirement> </requirements> <command detect_errors="exit_code"><