{"@id":"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl","retrievedFrom":{"repoUrl":"https://github.com/ncbi/pgap.git","branch":"6d04f5d65d1d4893706d9ae7e27341633333054f","path":"protein_alignment/wf_protein_alignment.cwl","packedId":null,"type":"GITHUB","rawUrl":"https://raw.githubusercontent.com/ncbi/pgap/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl","url":"https://github.com/ncbi/pgap/blob/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl"},"retrievedOn":"2026-09-05T10:51:19.324+00:00","lastCommit":"6d04f5d65d1d4893706d9ae7e27341633333054f","label":"Align reference proteins plane complete workflow","inputs":{"go":{"type":"Boolean[]"},"taxid":{"type":"Integer"},"asn_cache":{"type":"Directory"},"blastdb_dir":{"type":"Directory"},"gc_assembly":{"type":"File"},"compartments":{"type":"File"},"tax_sql_file":{"type":"File"},"naming_sqlite":{"type":"File"},"uniColl_asn_cache":{"type":"Directory"}},"outputs":{"align":{"type":"File"},"align_non_match":{"type":"File"},"universal_clusters":{"type":"File"}},"steps":{"cat":{"label":"file concatenation","run":"cat.cwl","runType":"COMMANDLINETOOL","sources":{"cat":{"sourceIDs":["Seed_Search_Compartments"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#cat/file_in_1":{"sourceIDs":["Seed_Protein_Alignments"]}}},"Get_Proteins":{"label":"Get Proteins, bacterial_prot_src","run":"bacterial_prot_src.cwl","runType":"COMMANDLINETOOL","sources":{"Get_Proteins":{"sourceIDs":["tax_sql_file"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Get_Proteins/taxid":{"sourceIDs":["taxid"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Get_Proteins/naming_sqlite":{"sourceIDs":["naming_sqlite"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Get_Proteins/uniColl_asn_cache":{"sourceIDs":["uniColl_asn_cache"]}}},"Sort_Seed_Hits":{"label":"Sort Seed Hits, align_sort","run":"align_sort.cwl","runType":"COMMANDLINETOOL","sources":{"Sort_Seed_Hits":{"sourceIDs":["uniColl_asn_cache"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Sort_Seed_Hits/asn_cache":{"sourceIDs":["asn_cache"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Sort_Seed_Hits/limit_mem":{"defaultVal":"\\\"1G\\\""},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Sort_Seed_Hits/blast_aligns":{"sourceIDs":["cat"]}}},"Compute_Gencode":{"label":"compute_gencode","run":"../progs/compute_gencode.cwl","runType":"COMMANDLINETOOL","sources":{"Compute_Gencode":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Compute_Gencode/taxid":{"sourceIDs":["taxid"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Compute_Gencode/taxon_db":{"sourceIDs":["tax_sql_file"]}}},"Compute_Gencode_int":{"run":"../progs/file2int.cwl","runType":"EXPRESSIONTOOL","sources":{"Compute_Gencode_int":{"sourceIDs":["Compute_Gencode"]}}},"Seed_Protein_Alignments":{"label":"Seed Protein Alignments I","run":"wf_seed_1.cwl","runType":"WORKFLOW","sources":{"Seed_Protein_Alignments":{"sourceIDs":["uniColl_asn_cache"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Seed_Protein_Alignments/seqids":{"sourceIDs":["Get_Proteins"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Seed_Protein_Alignments/asn_cache":{"sourceIDs":["asn_cache"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Seed_Protein_Alignments/db_gencode":{"sourceIDs":["Compute_Gencode_int"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Seed_Protein_Alignments/blastdb_dir":{"sourceIDs":["blastdb_dir"]}}},"Seed_Search_Compartments":{"label":"Seed Search Compartments","run":"wf_seed.cwl","runType":"WORKFLOW","sources":{"Seed_Search_Compartments":{"sourceIDs":["uniColl_asn_cache"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Seed_Search_Compartments/asn_cache":{"sourceIDs":["asn_cache"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Seed_Search_Compartments/db_gencode":{"sourceIDs":["Compute_Gencode_int"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Seed_Search_Compartments/compartments":{"sourceIDs":["compartments"]}}},"compart_filter_prosplign":{"label":"Filter Protein Seeds I; Find ProSplign Alignments I","run":"wf_compart_filter_prosplign.cwl","runType":"WORKFLOW","sources":{"compart_filter_prosplign":{"sourceIDs":["uniColl_asn_cache"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#compart_filter_prosplign/asn_cache":{"sourceIDs":["asn_cache"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#compart_filter_prosplign/seed_hits":{"sourceIDs":["Filter_Full_Coverage_Alignments_I"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#compart_filter_prosplign/gc_assembly":{"sourceIDs":["gc_assembly"]}}},"Filter_Protein_Alignments_I":{"label":"Filter Protein Alignments I","run":"wf_align_filter.cwl","runType":"WORKFLOW","sources":{"Filter_Protein_Alignments_I":{"sourceIDs":["compart_filter_prosplign"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Filter_Protein_Alignments_I/asn_cache":{"sourceIDs":["asn_cache"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Filter_Protein_Alignments_I/blast_full":{"sourceIDs":["Filter_Full_Coverage_Alignments_I"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Filter_Protein_Alignments_I/uniColl_asn_cache":{"sourceIDs":["uniColl_asn_cache"]}}},"Filter_Full_Coverage_Alignments_I":{"label":"Filter Full-Coverage Alignments I","run":"bacterial_protalign_filter.cwl","runType":"COMMANDLINETOOL","sources":{"Filter_Full_Coverage_Alignments_I":{"sourceIDs":["Sort_Seed_Hits"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Filter_Full_Coverage_Alignments_I/asn_cache":{"sourceIDs":["asn_cache"]},"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl#Filter_Full_Coverage_Alignments_I/uniColl_asn_cache":{"sourceIDs":["uniColl_asn_cache"]}}}},"cwltoolVersion":"3.1.20230201224320","visualisationDot":"digraph workflow {\n  graph [\n    bgcolor = \"#eeeeee\"\n    color = \"black\"\n    fontsize = \"10\"\n    labeljust = \"left\"\n    clusterrank = \"local\"\n    ranksep = \"0.22\"\n    nodesep = \"0.05\"\n  ]\n  node [\n    fontname = \"Helvetica\"\n    fontsize = \"10\"\n    fontcolor = \"black\"\n    shape = \"record\"\n    height = \"0\"\n    width = \"0\"\n    color = \"black\"\n    fillcolor = \"lightgoldenrodyellow\"\n    style = \"filled\"\n  ];\n  edge [\n    fontname=\"Helvetica\"\n    fontsize=\"8\"\n    fontcolor=\"black\"\n    color=\"black\"\n    arrowsize=\"0.7\"\n  ];\n  subgraph cluster_inputs {\n    rank = \"same\";\n    style = \"dashed\";\n    label = \"Workflow Inputs\";\n    \"uniColl_asn_cache\" [fillcolor=\"#94DDF4\",label=\"uniColl_asn_cache\"];\n    \"asn_cache\" [fillcolor=\"#94DDF4\",label=\"asn_cache\"];\n    \"gc_assembly\" [fillcolor=\"#94DDF4\",label=\"gc_assembly\"];\n    \"tax_sql_file\" [fillcolor=\"#94DDF4\",label=\"tax_sql_file\"];\n    \"compartments\" [fillcolor=\"#94DDF4\",label=\"compartments\"];\n    \"taxid\" [fillcolor=\"#94DDF4\",label=\"taxid\"];\n    \"blastdb_dir\" [fillcolor=\"#94DDF4\",label=\"blastdb_dir\"];\n    \"go\" [fillcolor=\"#94DDF4\",label=\"go\"];\n    \"naming_sqlite\" [fillcolor=\"#94DDF4\",label=\"naming_sqlite\"];\n  }\n  subgraph cluster_outputs {\n    rank = \"same\";\n    style = \"dashed\";\n    labelloc = \"b\";\n    label = \"Workflow Outputs\";\n    \"align\" [fillcolor=\"#94DDF4\",label=\"align\"];\n    \"align_non_match\" [fillcolor=\"#94DDF4\",label=\"align_non_match\"];\n    \"universal_clusters\" [fillcolor=\"#94DDF4\",label=\"universal_clusters\"];\n  }\n  \"Seed_Protein_Alignments\" [label=\"Seed Protein Alignments I\", fillcolor=\"#F3CEA1\"];\n  \"compart_filter_prosplign\" [label=\"Filter Protein Seeds I; Find ProSplign Alignments I\", fillcolor=\"#F3CEA1\"];\n  \"Get_Proteins\" [label=\"Get Proteins, bacterial_prot_src\"];\n  \"Seed_Search_Compartments\" [label=\"Seed Search Compartments\", fillcolor=\"#F3CEA1\"];\n  \"Filter_Full_Coverage_Alignments_I\" [label=\"Filter Full-Coverage Alignments I\"];\n  \"Sort_Seed_Hits\" [label=\"Sort Seed Hits, align_sort\"];\n  \"Compute_Gencode_int\" [label=\"Compute_Gencode_int\"];\n  \"Filter_Protein_Alignments_I\" [label=\"Filter Protein Alignments I\", fillcolor=\"#F3CEA1\"];\n  \"Compute_Gencode\" [label=\"compute_gencode\"];\n  \"cat\" [label=\"file concatenation\"];\n  \"uniColl_asn_cache\" -> \"Seed_Protein_Alignments\" [label=\"uniColl_asn_cache\"];\n  \"Get_Proteins\" -> \"Seed_Protein_Alignments\" [label=\"seqids\"];\n  \"Compute_Gencode_int\" -> \"Seed_Protein_Alignments\" [label=\"db_gencode\"];\n  \"blastdb_dir\" -> \"Seed_Protein_Alignments\" [label=\"blastdb_dir\"];\n  \"asn_cache\" -> \"Seed_Protein_Alignments\" [label=\"asn_cache\"];\n  \"uniColl_asn_cache\" -> \"compart_filter_prosplign\" [label=\"uniColl_asn_cache\"];\n  \"asn_cache\" -> \"compart_filter_prosplign\" [label=\"asn_cache\"];\n  \"gc_assembly\" -> \"compart_filter_prosplign\" [label=\"gc_assembly\"];\n  \"Filter_Full_Coverage_Alignments_I\" -> \"compart_filter_prosplign\" [label=\"seed_hits\"];\n  \"tax_sql_file\" -> \"Get_Proteins\" [label=\"tax_sql_file\"];\n  \"uniColl_asn_cache\" -> \"Get_Proteins\" [label=\"uniColl_asn_cache\"];\n  \"taxid\" -> \"Get_Proteins\" [label=\"taxid\"];\n  \"naming_sqlite\" -> \"Get_Proteins\" [label=\"naming_sqlite\"];\n  \"uniColl_asn_cache\" -> \"Seed_Search_Compartments\" [label=\"uniColl_asn_cache\"];\n  \"Compute_Gencode_int\" -> \"Seed_Search_Compartments\" [label=\"db_gencode\"];\n  \"compartments\" -> \"Seed_Search_Compartments\" [label=\"compartments\"];\n  \"asn_cache\" -> \"Seed_Search_Compartments\" [label=\"asn_cache\"];\n  \"Sort_Seed_Hits\" -> \"Filter_Full_Coverage_Alignments_I\" [label=\"sorted_seeds\"];\n  \"uniColl_asn_cache\" -> \"Filter_Full_Coverage_Alignments_I\" [label=\"uniColl_asn_cache\"];\n  \"asn_cache\" -> \"Filter_Full_Coverage_Alignments_I\" [label=\"asn_cache\"];\n  \"uniColl_asn_cache\" -> \"Sort_Seed_Hits\" [label=\"uniColl_asn_cache\"];\n  \"asn_cache\" -> \"Sort_Seed_Hits\" [label=\"asn_cache\"];\n  \"cat\" -> \"Sort_Seed_Hits\" [label=\"blast_aligns\"];\n  \"default1\" -> \"Sort_Seed_Hits\" [label=\"limit_mem\"];\n  \"default1\" [label=\"\\\"1G\\\"\", fillcolor=\"#D5AEFC\"];\n  \"Compute_Gencode\" -> \"Compute_Gencode_int\" [label=\"input\"];\n  \"compart_filter_prosplign\" -> \"Filter_Protein_Alignments_I\" [label=\"prosplign\"];\n  \"asn_cache\" -> \"Filter_Protein_Alignments_I\" [label=\"asn_cache\"];\n  \"uniColl_asn_cache\" -> \"Filter_Protein_Alignments_I\" [label=\"uniColl_asn_cache\"];\n  \"Filter_Full_Coverage_Alignments_I\" -> \"Filter_Protein_Alignments_I\" [label=\"blast_full\"];\n  \"default2\" -> \"Compute_Gencode\" [label=\"gencode\"];\n  \"default2\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"tax_sql_file\" -> \"Compute_Gencode\" [label=\"taxon_db\"];\n  \"taxid\" -> \"Compute_Gencode\" [label=\"taxid\"];\n  \"Seed_Search_Compartments\" -> \"cat\" [label=\"file_in_2\"];\n  \"Seed_Protein_Alignments\" -> \"cat\" [label=\"file_in_1\"];\n  \"Filter_Protein_Alignments_I\" -> \"align\";\n  \"Filter_Protein_Alignments_I\" -> \"align_non_match\";\n  \"Get_Proteins\" -> \"universal_clusters\";\n}\n","licenseLink":"https://raw.githubusercontent.com/ncbi/pgap/6d04f5d65d1d4893706d9ae7e27341633333054f/LICENSE.md","permalink":"https://w3id.org/cwl/view/git/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl","packed":false,"visualisationXdot":"/graph/xdot/github.com/ncbi/pgap/blob/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl","visualisationPng":"/graph/png/github.com/ncbi/pgap/blob/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl","visualisationSvg":"/graph/svg/github.com/ncbi/pgap/blob/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl","roBundle":"/robundle/github.com/ncbi/pgap/blob/6d04f5d65d1d4893706d9ae7e27341633333054f/protein_alignment/wf_protein_alignment.cwl","licenseName":"https://raw.githubusercontent.com/ncbi/pgap/6d04f5d65d1d4893706d9ae7e27341633333054f/LICENSE.md"}