{"@id":"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl","retrievedFrom":{"repoUrl":"https://github.com/ncbi/pgap.git","branch":"681555881cbdf95a1f4c29f6dc272eb6ca8e4d60","path":"bacterial_annot/wf_bacterial_annot_pass2.cwl","packedId":null,"type":"GITHUB","rawUrl":"https://raw.githubusercontent.com/ncbi/pgap/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl","url":"https://github.com/ncbi/pgap/blob/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl"},"retrievedOn":"2026-08-31T15:00:47.048+00:00","lastCommit":"681555881cbdf95a1f4c29f6dc272eb6ca8e4d60","label":"Bacterial Annotation, pass 2, blastp-based functional annotation (first pass)","inputs":{"lds2":{"label":"Extract ORF Proteins/lds2","type":"File"},"taxid":{"type":"Integer"},"blastdb":{"label":"Input blastdb databases","type":"String[]"},"plasmids":{"label":"Prepare_Unannotated_Sequences/plasmids","type":"File"},"proteins":{"label":"Extract ORF Proteins/proteins","type":"File"},"raw_seqs":{"label":"Prepare_Unannotated_Sequences/sequences","type":"File"},"taxon_db":{"type":"File"},"annotation":{"label":"Get ORFs/outseq","type":"File"},"genus_list":{"type":"Integer[]"},"prot_ids_A":{"label":"Extract ORF Proteins/prot_ids","type":"File"},"prot_ids_B1":{"label":"Get off-frame ORFs/prot_ids","type":"File"},"prot_ids_B2":{"label":"AntiFam tainted proteins I/oseqids","type":"File"},"unicoll_cache":{"type":"Directory"},"blast_rules_db":{"label":"parameter to store the literal 'blast_rules_db'","type":"String"},"sequence_cache":{"type":"Directory"},"blast_hits_cache":{"type":"File (Optional)"},"identification_db_dir":{"label":"Create identification BLASTdb","type":"Directory"},"scatter_gather_nchunks":{"type":"String"}},"outputs":{"aligns":{"label":"goes to protein_alignment/Seed Search Compartments/compartments","type":"File"}},"steps":{"Map_Naming_Hits":{"label":"bacterial_hit_mapping","run":"bacterial_hit_mapping.cwl","runType":"COMMANDLINETOOL","sources":{"Map_Naming_Hits":{"sourceIDs":["Find_Struct_Protein_Hits_for_Plasmids"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Map_Naming_Hits/hmm_hits":{"sourceIDs":["Find_Naming_Protein_Hits_for_struct"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Map_Naming_Hits/proteins":{"sourceIDs":["proteins"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Map_Naming_Hits/asn_cache":{"sourceIDs":["sequence_cache"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Map_Naming_Hits/sequences":{"sourceIDs":["annotation"]}}},"Remove_off_frame_ORFs":{"label":"set_operation","run":"../progs/set_operation.cwl","runType":"COMMANDLINETOOL","sources":{"Remove_off_frame_ORFs":{"sourceIDs":["prot_ids_A"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Remove_off_frame_ORFs/B":{"sourceIDs":["prot_ids_B1"]}}},"Separate_Plasmid_ORFs":{"label":"separate_plasmid_orfs","run":"../progs/separate_plasmid_orfs.cwl","runType":"COMMANDLINETOOL","sources":{"Separate_Plasmid_ORFs":{"sourceIDs":["plasmids"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Separate_Plasmid_ORFs/orf_ids":{"sourceIDs":["Remove_off_frame_ORFs"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Separate_Plasmid_ORFs/raw_seqs":{"sourceIDs":["raw_seqs"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Separate_Plasmid_ORFs/orf_entries":{"sourceIDs":["annotation"]}}},"Find_Naming_Protein_Hits_for_struct":{"label":"blastp_wnode_naming","run":"../task_types/tt_blastp_wnode_naming.cwl","runType":"WORKFLOW","sources":{"Find_Naming_Protein_Hits_for_struct":{"sourceIDs":["taxon_db"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_for_struct/ids":{"sourceIDs":["Separate_Plasmid_ORFs"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_for_struct/lds2":{"sourceIDs":["lds2"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_for_struct/taxid":{"sourceIDs":["taxid"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_for_struct/blastdb":{"sourceIDs":["blastdb"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_for_struct/proteins":{"sourceIDs":["proteins"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_for_struct/asn_cache":{"sourceIDs":["sequence_cache"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_for_struct/genus_list":{"sourceIDs":["genus_list"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_for_struct/blastdb_dir":{"sourceIDs":["identification_db_dir"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_for_struct/blast_hits_cache":{"sourceIDs":["blast_hits_cache"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_for_struct/scatter_gather_nchunks":{"sourceIDs":["scatter_gather_nchunks"]}}},"Find_Struct_Protein_Hits_for_Plasmids":{"label":"blastp_wnode_struct","run":"../task_types/tt_blastp_wnode_struct.cwl","runType":"WORKFLOW","sources":{"Find_Struct_Protein_Hits_for_Plasmids":{"sourceIDs":["taxon_db"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Struct_Protein_Hits_for_Plasmids/ids":{"sourceIDs":["Separate_Plasmid_ORFs"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Struct_Protein_Hits_for_Plasmids/lds2":{"sourceIDs":["lds2"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Struct_Protein_Hits_for_Plasmids/taxid":{"sourceIDs":["taxid"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Struct_Protein_Hits_for_Plasmids/proteins":{"sourceIDs":["proteins"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Struct_Protein_Hits_for_Plasmids/asn_cache":{"sourceIDs":["sequence_cache"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Struct_Protein_Hits_for_Plasmids/genus_list":{"sourceIDs":["genus_list"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Struct_Protein_Hits_for_Plasmids/blastdb_dir":{"sourceIDs":["identification_db_dir"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Struct_Protein_Hits_for_Plasmids/blast_hits_cache":{"sourceIDs":["blast_hits_cache"]},"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Struct_Protein_Hits_for_Plasmids/scatter_gather_nchunks":{"sourceIDs":["scatter_gather_nchunks"]}}}},"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    \"taxon_db\" [fillcolor=\"#94DDF4\",label=\"taxon_db\"];\n    \"unicoll_cache\" [fillcolor=\"#94DDF4\",label=\"unicoll_cache\"];\n    \"sequence_cache\" [fillcolor=\"#94DDF4\",label=\"sequence_cache\"];\n    \"scatter_gather_nchunks\" [fillcolor=\"#94DDF4\",label=\"scatter_gather_nchunks\"];\n    \"taxid\" [fillcolor=\"#94DDF4\",label=\"taxid\"];\n    \"proteins\" [fillcolor=\"#94DDF4\",label=\"Extract ORF Proteins/proteins\"];\n    \"blast_rules_db\" [fillcolor=\"#94DDF4\",label=\"parameter to store the literal 'blast_rules_db'\"];\n    \"identification_db_dir\" [fillcolor=\"#94DDF4\",label=\"Create identification BLASTdb\"];\n    \"plasmids\" [fillcolor=\"#94DDF4\",label=\"Prepare_Unannotated_Sequences/plasmids\"];\n    \"blast_hits_cache\" [fillcolor=\"#94DDF4\",label=\"blast_hits_cache\"];\n    \"raw_seqs\" [fillcolor=\"#94DDF4\",label=\"Prepare_Unannotated_Sequences/sequences\"];\n    \"prot_ids_A\" [fillcolor=\"#94DDF4\",label=\"Extract ORF Proteins/prot_ids\"];\n    \"genus_list\" [fillcolor=\"#94DDF4\",label=\"genus_list\"];\n    \"prot_ids_B2\" [fillcolor=\"#94DDF4\",label=\"AntiFam tainted proteins I/oseqids\"];\n    \"prot_ids_B1\" [fillcolor=\"#94DDF4\",label=\"Get off-frame ORFs/prot_ids\"];\n    \"annotation\" [fillcolor=\"#94DDF4\",label=\"Get ORFs/outseq\"];\n    \"lds2\" [fillcolor=\"#94DDF4\",label=\"Extract ORF Proteins/lds2\"];\n    \"blastdb\" [fillcolor=\"#94DDF4\",label=\"Input blastdb databases\"];\n  }\n  subgraph cluster_outputs {\n    rank = \"same\";\n    style = \"dashed\";\n    labelloc = \"b\";\n    label = \"Workflow Outputs\";\n    \"aligns\" [fillcolor=\"#94DDF4\",label=\"goes to protein_alignment/Seed Search Compartments/compartments\"];\n  }\n  \"Separate_Plasmid_ORFs\" [label=\"separate_plasmid_orfs\"];\n  \"Find_Naming_Protein_Hits_for_struct\" [label=\"blastp_wnode_naming\", fillcolor=\"#F3CEA1\"];\n  \"Remove_off_frame_ORFs\" [label=\"set_operation\"];\n  \"Find_Struct_Protein_Hits_for_Plasmids\" [label=\"blastp_wnode_struct\", fillcolor=\"#F3CEA1\"];\n  \"Map_Naming_Hits\" [label=\"bacterial_hit_mapping\"];\n  \"plasmids\" -> \"Separate_Plasmid_ORFs\" [label=\"plasmids\"];\n  \"Remove_off_frame_ORFs\" -> \"Separate_Plasmid_ORFs\" [label=\"orf_ids\"];\n  \"raw_seqs\" -> \"Separate_Plasmid_ORFs\" [label=\"raw_seqs\"];\n  \"annotation\" -> \"Separate_Plasmid_ORFs\" [label=\"orf_entries\"];\n  \"taxon_db\" -> \"Find_Naming_Protein_Hits_for_struct\" [label=\"taxon_db\"];\n  \"unicoll_cache\" -> \"Find_Naming_Protein_Hits_for_struct\" [label=\"asn_cache\"];\n  \"sequence_cache\" -> \"Find_Naming_Protein_Hits_for_struct\" [label=\"asn_cache\"];\n  \"scatter_gather_nchunks\" -> \"Find_Naming_Protein_Hits_for_struct\" [label=\"scatter_gather_nchunks\"];\n  \"taxid\" -> \"Find_Naming_Protein_Hits_for_struct\" [label=\"taxid\"];\n  \"blast_rules_db\" -> \"Find_Naming_Protein_Hits_for_struct\" [label=\"blastdb\"];\n  \"blastdb\" -> \"Find_Naming_Protein_Hits_for_struct\" [label=\"blastdb\"];\n  \"genus_list\" -> \"Find_Naming_Protein_Hits_for_struct\" [label=\"genus_list\"];\n  \"Separate_Plasmid_ORFs\" -> \"Find_Naming_Protein_Hits_for_struct\" [label=\"ids\"];\n  \"blast_hits_cache\" -> \"Find_Naming_Protein_Hits_for_struct\" [label=\"blast_hits_cache\"];\n  \"lds2\" -> \"Find_Naming_Protein_Hits_for_struct\" [label=\"lds2\"];\n  \"identification_db_dir\" -> \"Find_Naming_Protein_Hits_for_struct\" [label=\"blastdb_dir\"];\n  \"proteins\" -> \"Find_Naming_Protein_Hits_for_struct\" [label=\"proteins\"];\n  \"prot_ids_A\" -> \"Remove_off_frame_ORFs\" [label=\"A\"];\n  \"prot_ids_B2\" -> \"Remove_off_frame_ORFs\" [label=\"B\"];\n  \"prot_ids_B1\" -> \"Remove_off_frame_ORFs\" [label=\"B\"];\n  \"taxon_db\" -> \"Find_Struct_Protein_Hits_for_Plasmids\" [label=\"taxon_db\"];\n  \"proteins\" -> \"Find_Struct_Protein_Hits_for_Plasmids\" [label=\"proteins\"];\n  \"blast_hits_cache\" -> \"Find_Struct_Protein_Hits_for_Plasmids\" [label=\"blast_hits_cache\"];\n  \"unicoll_cache\" -> \"Find_Struct_Protein_Hits_for_Plasmids\" [label=\"asn_cache\"];\n  \"sequence_cache\" -> \"Find_Struct_Protein_Hits_for_Plasmids\" [label=\"asn_cache\"];\n  \"genus_list\" -> \"Find_Struct_Protein_Hits_for_Plasmids\" [label=\"genus_list\"];\n  \"lds2\" -> \"Find_Struct_Protein_Hits_for_Plasmids\" [label=\"lds2\"];\n  \"taxid\" -> \"Find_Struct_Protein_Hits_for_Plasmids\" [label=\"taxid\"];\n  \"Separate_Plasmid_ORFs\" -> \"Find_Struct_Protein_Hits_for_Plasmids\" [label=\"ids\"];\n  \"scatter_gather_nchunks\" -> \"Find_Struct_Protein_Hits_for_Plasmids\" [label=\"scatter_gather_nchunks\"];\n  \"identification_db_dir\" -> \"Find_Struct_Protein_Hits_for_Plasmids\" [label=\"blastdb_dir\"];\n  \"Find_Struct_Protein_Hits_for_Plasmids\" -> \"Map_Naming_Hits\" [label=\"hmm_hits\"];\n  \"Find_Naming_Protein_Hits_for_struct\" -> \"Map_Naming_Hits\" [label=\"hmm_hits\"];\n  \"annotation\" -> \"Map_Naming_Hits\" [label=\"sequences\"];\n  \"unicoll_cache\" -> \"Map_Naming_Hits\" [label=\"asn_cache\"];\n  \"sequence_cache\" -> \"Map_Naming_Hits\" [label=\"asn_cache\"];\n  \"proteins\" -> \"Map_Naming_Hits\" [label=\"proteins\"];\n  \"Map_Naming_Hits\" -> \"aligns\";\n}\n","licenseLink":"https://raw.githubusercontent.com/ncbi/pgap/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/LICENSE.md","permalink":"https://w3id.org/cwl/view/git/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl","packed":false,"visualisationXdot":"/graph/xdot/github.com/ncbi/pgap/blob/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl","visualisationPng":"/graph/png/github.com/ncbi/pgap/blob/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl","visualisationSvg":"/graph/svg/github.com/ncbi/pgap/blob/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl","roBundle":"/robundle/github.com/ncbi/pgap/blob/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/bacterial_annot/wf_bacterial_annot_pass2.cwl","licenseName":"https://raw.githubusercontent.com/ncbi/pgap/681555881cbdf95a1f4c29f6dc272eb6ca8e4d60/LICENSE.md"}