{"@id":"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl","retrievedFrom":{"repoUrl":"https://github.com/ncbi/pgap.git","branch":"7f857f7f2d7c080d27c775b67a6d6f7d94bce31f","path":"bacterial_annot/wf_bacterial_annot_pass2.cwl","packedId":null,"type":"GITHUB","rawUrl":"https://raw.githubusercontent.com/ncbi/pgap/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl","url":"https://github.com/ncbi/pgap/blob/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl"},"retrievedOn":"2026-09-22T04:21:52.083+00:00","lastCommit":"7f857f7f2d7c080d27c775b67a6d6f7d94bce31f","label":"Bacterial Annotation, pass 2, blastp-based functional annotation (first pass)","inputs":{"lds2":{"label":"Extract ORF Proteins/lds2","type":"File"},"proteins":{"label":"Extract ORF Proteins/proteins","type":"File"},"annotation":{"label":"Get ORFs/outseq","type":"File"},"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":"Name of blast_rules_db","type":"String"},"sequence_cache":{"type":"Directory"},"blast_rules_db_dir":{"label":"Get BLAST Rules db const","type":"Directory"},"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":"Map HMM Hits","run":"bacterial_hit_mapping.cwl","runType":"COMMANDLINETOOL","sources":{"Map_Naming_Hits":{"sourceIDs":["annotation"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Map_Naming_Hits/hmm_hits":{"sourceIDs":["Find_Naming_Protein_Hits_I"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Map_Naming_Hits/proteins":{"sourceIDs":["proteins"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Map_Naming_Hits/align_fmt":{"defaultVal":"\\\"seq-align-set\\\""},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Map_Naming_Hits/asn_cache":{"sourceIDs":["unicoll_cache"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Map_Naming_Hits/nogenbank":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Map_Naming_Hits/no_compart":{"defaultVal":"false"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Map_Naming_Hits/expansion_ratio":{"defaultVal":"1.1e+00"}}},"Remove_off_frame_ORFs":{"label":"set_operation","run":"../progs/set_operation.cwl","runType":"COMMANDLINETOOL","sources":{"Remove_off_frame_ORFs":{"defaultVal":"\\\"-\\\""},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Remove_off_frame_ORFs/A":{"sourceIDs":["prot_ids_A"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Remove_off_frame_ORFs/B":{"sourceIDs":["prot_ids_B1"]}}},"Find_Naming_Protein_Hits_I":{"label":"blastp_wnode_naming","run":"../task_types/tt_blastp_wnode_naming.cwl","runType":"WORKFLOW","sources":{"Find_Naming_Protein_Hits_I":{"sourceIDs":["proteins"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/ids":{"sourceIDs":["Remove_off_frame_ORFs"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/seg":{"defaultVal":"\\\"no\\\""},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/lds2":{"sourceIDs":["lds2"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/ofmt":{"defaultVal":"\\\"asn-binary\\\""},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/dbsize":{"defaultVal":"\\\"6000000000\\\""},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/evalue":{"defaultVal":"1e-01"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/blastdb":{"defaultVal":"\\\"blast_rules_db\\\""},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/compart":{"defaultVal":"false"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/affinity":{"defaultVal":"\\\"subject\\\""},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/max_jobs":{"defaultVal":"1"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/no_merge":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/asn_cache":{"sourceIDs":["unicoll_cache"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/nogenbank":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/threshold":{"defaultVal":"21"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/word_size":{"defaultVal":"6"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/blastdb_dir":{"sourceIDs":["blast_rules_db_dir"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/align_filter":{"defaultVal":"\\\"score>0 && pct_identity_gapopen_only > 35\\\""},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/top_by_score":{"defaultVal":"10"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/extra_coverage":{"defaultVal":"20"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/max_target_seqs":{"defaultVal":"50"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/comp_based_stats":{"defaultVal":"\\\"F\\\""},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/max_batch_length":{"defaultVal":"10000"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/allow_intersection":{"defaultVal":"false"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl#Find_Naming_Protein_Hits_I/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    \"sequence_cache\" [fillcolor=\"#94DDF4\",label=\"sequence_cache\"];\n    \"unicoll_cache\" [fillcolor=\"#94DDF4\",label=\"unicoll_cache\"];\n    \"blast_rules_db\" [fillcolor=\"#94DDF4\",label=\"Name of blast_rules_db\"];\n    \"lds2\" [fillcolor=\"#94DDF4\",label=\"Extract ORF Proteins/lds2\"];\n    \"scatter_gather_nchunks\" [fillcolor=\"#94DDF4\",label=\"scatter_gather_nchunks\"];\n    \"prot_ids_B2\" [fillcolor=\"#94DDF4\",label=\"AntiFam tainted proteins I/oseqids\"];\n    \"prot_ids_A\" [fillcolor=\"#94DDF4\",label=\"Extract ORF Proteins/prot_ids\"];\n    \"annotation\" [fillcolor=\"#94DDF4\",label=\"Get ORFs/outseq\"];\n    \"prot_ids_B1\" [fillcolor=\"#94DDF4\",label=\"Get off-frame ORFs/prot_ids\"];\n    \"identification_db_dir\" [fillcolor=\"#94DDF4\",label=\"Create identification BLASTdb\"];\n    \"blast_rules_db_dir\" [fillcolor=\"#94DDF4\",label=\"Get BLAST Rules db const\"];\n    \"proteins\" [fillcolor=\"#94DDF4\",label=\"Extract ORF Proteins/proteins\"];\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  \"Map_Naming_Hits\" [label=\"Map HMM Hits\"];\n  \"Remove_off_frame_ORFs\" [label=\"set_operation\"];\n  \"Find_Naming_Protein_Hits_I\" [label=\"blastp_wnode_naming\", fillcolor=\"#F3CEA1\"];\n  \"annotation\" -> \"Map_Naming_Hits\" [label=\"sequences\"];\n  \"default1\" -> \"Map_Naming_Hits\" [label=\"no_compart\"];\n  \"default1\" [label=\"false\", fillcolor=\"#D5AEFC\"];\n  \"default2\" -> \"Map_Naming_Hits\" [label=\"align_fmt\"];\n  \"default2\" [label=\"\\\"seq-align-set\\\"\", fillcolor=\"#D5AEFC\"];\n  \"sequence_cache\" -> \"Map_Naming_Hits\" [label=\"asn_cache\"];\n  \"unicoll_cache\" -> \"Map_Naming_Hits\" [label=\"asn_cache\"];\n  \"proteins\" -> \"Map_Naming_Hits\" [label=\"proteins\"];\n  \"default3\" -> \"Map_Naming_Hits\" [label=\"nogenbank\"];\n  \"default3\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"Find_Naming_Protein_Hits_I\" -> \"Map_Naming_Hits\" [label=\"hmm_hits\"];\n  \"default4\" -> \"Map_Naming_Hits\" [label=\"expansion_ratio\"];\n  \"default4\" [label=\"1.1e+00\", fillcolor=\"#D5AEFC\"];\n  \"default5\" -> \"Remove_off_frame_ORFs\" [label=\"operation\"];\n  \"default5\" [label=\"\\\"-\\\"\", fillcolor=\"#D5AEFC\"];\n  \"prot_ids_B2\" -> \"Remove_off_frame_ORFs\" [label=\"B\"];\n  \"prot_ids_B1\" -> \"Remove_off_frame_ORFs\" [label=\"B\"];\n  \"prot_ids_A\" -> \"Remove_off_frame_ORFs\" [label=\"A\"];\n  \"proteins\" -> \"Find_Naming_Protein_Hits_I\" [label=\"proteins\"];\n  \"default6\" -> \"Find_Naming_Protein_Hits_I\" [label=\"seg\"];\n  \"default6\" [label=\"\\\"no\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default7\" -> \"Find_Naming_Protein_Hits_I\" [label=\"affinity\"];\n  \"default7\" [label=\"\\\"subject\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default8\" -> \"Find_Naming_Protein_Hits_I\" [label=\"allow_intersection\"];\n  \"default8\" [label=\"false\", fillcolor=\"#D5AEFC\"];\n  \"default9\" -> \"Find_Naming_Protein_Hits_I\" [label=\"comp_based_stats\"];\n  \"default9\" [label=\"\\\"F\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default10\" -> \"Find_Naming_Protein_Hits_I\" [label=\"max_jobs\"];\n  \"default10\" [label=\"1\", fillcolor=\"#D5AEFC\"];\n  \"default11\" -> \"Find_Naming_Protein_Hits_I\" [label=\"blastdb\"];\n  \"default11\" [label=\"\\\"blastdb\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default12\" -> \"Find_Naming_Protein_Hits_I\" [label=\"blastdb\"];\n  \"default12\" [label=\"\\\"blast_rules_db\\\"\", fillcolor=\"#D5AEFC\"];\n  \"sequence_cache\" -> \"Find_Naming_Protein_Hits_I\" [label=\"asn_cache\"];\n  \"unicoll_cache\" -> \"Find_Naming_Protein_Hits_I\" [label=\"asn_cache\"];\n  \"default13\" -> \"Find_Naming_Protein_Hits_I\" [label=\"nogenbank\"];\n  \"default13\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"default14\" -> \"Find_Naming_Protein_Hits_I\" [label=\"no_merge\"];\n  \"default14\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"default15\" -> \"Find_Naming_Protein_Hits_I\" [label=\"top_by_score\"];\n  \"default15\" [label=\"10\", fillcolor=\"#D5AEFC\"];\n  \"Remove_off_frame_ORFs\" -> \"Find_Naming_Protein_Hits_I\" [label=\"ids\"];\n  \"default16\" -> \"Find_Naming_Protein_Hits_I\" [label=\"max_target_seqs\"];\n  \"default16\" [label=\"50\", fillcolor=\"#D5AEFC\"];\n  \"default17\" -> \"Find_Naming_Protein_Hits_I\" [label=\"word_size\"];\n  \"default17\" [label=\"6\", fillcolor=\"#D5AEFC\"];\n  \"default18\" -> \"Find_Naming_Protein_Hits_I\" [label=\"threshold\"];\n  \"default18\" [label=\"21\", fillcolor=\"#D5AEFC\"];\n  \"default19\" -> \"Find_Naming_Protein_Hits_I\" [label=\"max_batch_length\"];\n  \"default19\" [label=\"10000\", fillcolor=\"#D5AEFC\"];\n  \"identification_db_dir\" -> \"Find_Naming_Protein_Hits_I\" [label=\"blastdb_dir\"];\n  \"blast_rules_db_dir\" -> \"Find_Naming_Protein_Hits_I\" [label=\"blastdb_dir\"];\n  \"scatter_gather_nchunks\" -> \"Find_Naming_Protein_Hits_I\" [label=\"scatter_gather_nchunks\"];\n  \"default20\" -> \"Find_Naming_Protein_Hits_I\" [label=\"align_filter\"];\n  \"default20\" [label=\"\\\"score>0 && pct_identity_gapopen_only > 35\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default21\" -> \"Find_Naming_Protein_Hits_I\" [label=\"dbsize\"];\n  \"default21\" [label=\"\\\"6000000000\\\"\", fillcolor=\"#D5AEFC\"];\n  \"lds2\" -> \"Find_Naming_Protein_Hits_I\" [label=\"lds2\"];\n  \"default22\" -> \"Find_Naming_Protein_Hits_I\" [label=\"extra_coverage\"];\n  \"default22\" [label=\"20\", fillcolor=\"#D5AEFC\"];\n  \"default23\" -> \"Find_Naming_Protein_Hits_I\" [label=\"compart\"];\n  \"default23\" [label=\"false\", fillcolor=\"#D5AEFC\"];\n  \"default24\" -> \"Find_Naming_Protein_Hits_I\" [label=\"evalue\"];\n  \"default24\" [label=\"1e-01\", fillcolor=\"#D5AEFC\"];\n  \"default25\" -> \"Find_Naming_Protein_Hits_I\" [label=\"ofmt\"];\n  \"default25\" [label=\"\\\"asn-binary\\\"\", fillcolor=\"#D5AEFC\"];\n  \"Map_Naming_Hits\" -> \"aligns\";\n}\n","licenseLink":"https://raw.githubusercontent.com/ncbi/pgap/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/LICENSE.md","permalink":"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl","packed":false,"visualisationXdot":"/graph/xdot/github.com/ncbi/pgap/blob/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl","visualisationPng":"/graph/png/github.com/ncbi/pgap/blob/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl","visualisationSvg":"/graph/svg/github.com/ncbi/pgap/blob/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl","roBundle":"/robundle/github.com/ncbi/pgap/blob/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass2.cwl","licenseName":"https://raw.githubusercontent.com/ncbi/pgap/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/LICENSE.md"}