{"@id":"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl","retrievedFrom":{"repoUrl":"https://github.com/ncbi/pgap.git","branch":"7f857f7f2d7c080d27c775b67a6d6f7d94bce31f","path":"bacterial_annot/wf_bacterial_annot_pass1.cwl","packedId":null,"type":"GITHUB","rawUrl":"https://raw.githubusercontent.com/ncbi/pgap/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl","url":"https://github.com/ncbi/pgap/blob/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl"},"retrievedOn":"2026-09-21T06:28:28.475+00:00","lastCommit":"7f857f7f2d7c080d27c775b67a6d6f7d94bce31f","label":"Bacterial Annotation, pass 1, genemark training, by HMMs (first pass)","inputs":{"inseq":{"type":"File"},"hmm_path":{"type":"Directory"},"hmms_tab":{"type":"File"},"asn_cache":{"type":"Directory"},"nogenbank":{"type":"Boolean"},"thresholds":{"type":"File"},"trna_annots":{"type":"File"},"ncrna_annots":{"type":"File"},"genemark_path":{"type":"Directory"},"uniColl_cache":{"type":"Directory"},"Execute_CRISPRs_annots":{"type":"File"},"scatter_gather_nchunks":{"type":"String"},"Generate_16S_rRNA_Annotation_annotation":{"type":"File"},"Generate_23S_rRNA_Annotation_annotation":{"type":"File"},"Post_process_CMsearch_annotations_annots_5S":{"type":"File"}},"outputs":{"lds2":{"type":"File"},"aligns":{"type":"File"},"seqids":{"type":"File"},"outseqs":{"type":"File"},"hmm_hits":{"type":"File"},"prot_ids":{"type":"File"},"proteins":{"type":"File"},"annotation":{"type":"File"},"out_hmm_params":{"type":"File"},"protein_aligns":{"type":"File"}},"steps":{"Get_ORFs":{"label":"gp_getorf","run":"../progs/gp_getorf.cwl","runType":"COMMANDLINETOOL","sources":{"Get_ORFs":{"sourceIDs":["inseq"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Get_ORFs/asn_cache":{"sourceIDs":["asn_cache"]}}},"Map_HMM_Hits":{"label":"Map HMM Hits","run":"bacterial_hit_mapping.cwl","runType":"COMMANDLINETOOL","sources":{"Map_HMM_Hits":{"sourceIDs":["Search_All_HMMs_I"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Map_HMM_Hits/align_fmt":{"defaultVal":"\\\"seq-align\\\""},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Map_HMM_Hits/asn_cache":{"sourceIDs":["uniColl_cache"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Map_HMM_Hits/nogenbank":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Map_HMM_Hits/seq_cache":{"sourceIDs":["asn_cache"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Map_HMM_Hits/sequences":{"sourceIDs":["Get_ORFs"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Map_HMM_Hits/no_compart":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Map_HMM_Hits/unicoll_cache":{"sourceIDs":["uniColl_cache"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Map_HMM_Hits/expansion_ratio":{"defaultVal":"0e+00"}}},"Search_All_HMMs_I":{"run":"../task_types/tt_hmmsearch_wnode.cwl","runType":"WORKFLOW","sources":{"Search_All_HMMs_I":{"sourceIDs":["hmms_tab"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Search_All_HMMs_I/lds2":{"sourceIDs":["Extract_ORF_Proteins"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Search_All_HMMs_I/seqids":{"sourceIDs":["Extract_ORF_Proteins"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Search_All_HMMs_I/hmm_path":{"sourceIDs":["hmm_path"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Search_All_HMMs_I/proteins":{"sourceIDs":["Extract_ORF_Proteins"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Search_All_HMMs_I/asn_cache":{"sourceIDs":["asn_cache"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Search_All_HMMs_I/scatter_gather_nchunks":{"sourceIDs":["scatter_gather_nchunks"]}}},"Get_off_frame_ORFs":{"label":"Get off-frame ORFs","run":"get_off_frame_orfs.cwl","runType":"COMMANDLINETOOL","sources":{"Get_off_frame_ORFs":{"sourceIDs":["Get_ORFs"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Get_off_frame_ORFs/aligns":{"sourceIDs":["Map_HMM_Hits"]}}},"Extract_ORF_Proteins":{"label":"protein_extract","run":"../progs/protein_extract.cwl","runType":"WORKFLOW","sources":{"Extract_ORF_Proteins":{"sourceIDs":["Get_ORFs"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Extract_ORF_Proteins/nogenbank":{"sourceIDs":["nogenbank"]}}},"Run_GeneMark_Training":{"label":"genemark_training","run":"../progs/genemark_training.cwl","runType":"COMMANDLINETOOL","sources":{"Run_GeneMark_Training":{"sourceIDs":["genemark_path"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Run_GeneMark_Training/thr":{"sourceIDs":["thresholds"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Run_GeneMark_Training/asn_cache":{"sourceIDs":["uniColl_cache"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Run_GeneMark_Training/nogenbank":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Run_GeneMark_Training/sequences":{"sourceIDs":["inseq"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Run_GeneMark_Training/min_seq_len":{"defaultVal":"200"},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Run_GeneMark_Training/tmp_dir_name":{"defaultVal":"\\\"workdir\\\""},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Run_GeneMark_Training/marked_annotation_name":{"defaultVal":"\\\"marked-annotation.asn\\\""},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Run_GeneMark_Training/preliminary_models_name":{"defaultVal":"\\\"preliminary-models.asn\\\""}}},"Resolve_Annotation_Conflicts":{"label":"bacterial_resolve_conflicts","run":"../progs/bacterial_resolve_conflicts.cwl","runType":"COMMANDLINETOOL","sources":{"Resolve_Annotation_Conflicts":{"sourceIDs":["asn_cache"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Resolve_Annotation_Conflicts/thr":{"sourceIDs":["thresholds"]},"https://w3id.org/cwl/view/git/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl#Resolve_Annotation_Conflicts/features":{"sourceIDs":["ncrna_annots"]}}}},"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    \"Execute_CRISPRs_annots\" [fillcolor=\"#94DDF4\",label=\"Execute_CRISPRs_annots\"];\n    \"hmms_tab\" [fillcolor=\"#94DDF4\",label=\"hmms_tab\"];\n    \"Generate_16S_rRNA_Annotation_annotation\" [fillcolor=\"#94DDF4\",label=\"Generate_16S_rRNA_Annotation_annotation\"];\n    \"asn_cache\" [fillcolor=\"#94DDF4\",label=\"asn_cache\"];\n    \"scatter_gather_nchunks\" [fillcolor=\"#94DDF4\",label=\"scatter_gather_nchunks\"];\n    \"Generate_23S_rRNA_Annotation_annotation\" [fillcolor=\"#94DDF4\",label=\"Generate_23S_rRNA_Annotation_annotation\"];\n    \"hmm_path\" [fillcolor=\"#94DDF4\",label=\"hmm_path\"];\n    \"genemark_path\" [fillcolor=\"#94DDF4\",label=\"genemark_path\"];\n    \"nogenbank\" [fillcolor=\"#94DDF4\",label=\"nogenbank\"];\n    \"Post_process_CMsearch_annotations_annots_5S\" [fillcolor=\"#94DDF4\",label=\"Post_process_CMsearch_annotations_annots_5S\"];\n    \"thresholds\" [fillcolor=\"#94DDF4\",label=\"thresholds\"];\n    \"trna_annots\" [fillcolor=\"#94DDF4\",label=\"trna_annots\"];\n    \"ncrna_annots\" [fillcolor=\"#94DDF4\",label=\"ncrna_annots\"];\n    \"uniColl_cache\" [fillcolor=\"#94DDF4\",label=\"uniColl_cache\"];\n    \"inseq\" [fillcolor=\"#94DDF4\",label=\"inseq\"];\n  }\n  subgraph cluster_outputs {\n    rank = \"same\";\n    style = \"dashed\";\n    labelloc = \"b\";\n    label = \"Workflow Outputs\";\n    \"protein_aligns\" [fillcolor=\"#94DDF4\",label=\"protein_aligns\"];\n    \"proteins\" [fillcolor=\"#94DDF4\",label=\"proteins\"];\n    \"aligns\" [fillcolor=\"#94DDF4\",label=\"aligns\"];\n    \"hmm_hits\" [fillcolor=\"#94DDF4\",label=\"hmm_hits\"];\n    \"outseqs\" [fillcolor=\"#94DDF4\",label=\"outseqs\"];\n    \"seqids\" [fillcolor=\"#94DDF4\",label=\"seqids\"];\n    \"lds2\" [fillcolor=\"#94DDF4\",label=\"lds2\"];\n    \"out_hmm_params\" [fillcolor=\"#94DDF4\",label=\"out_hmm_params\"];\n    \"prot_ids\" [fillcolor=\"#94DDF4\",label=\"prot_ids\"];\n    \"annotation\" [fillcolor=\"#94DDF4\",label=\"annotation\"];\n  }\n  \"Run_GeneMark_Training\" [label=\"genemark_training\"];\n  \"Get_off_frame_ORFs\" [label=\"Get off-frame ORFs\"];\n  \"Map_HMM_Hits\" [label=\"Map HMM Hits\"];\n  \"Get_ORFs\" [label=\"gp_getorf\"];\n  \"Search_All_HMMs_I\" [label=\"Search_All_HMMs_I\", fillcolor=\"#F3CEA1\"];\n  \"Extract_ORF_Proteins\" [label=\"protein_extract\", fillcolor=\"#F3CEA1\"];\n  \"Resolve_Annotation_Conflicts\" [label=\"bacterial_resolve_conflicts\"];\n  \"genemark_path\" -> \"Run_GeneMark_Training\" [label=\"genemark_path\"];\n  \"default1\" -> \"Run_GeneMark_Training\" [label=\"preliminary_models_name\"];\n  \"default1\" [label=\"\\\"preliminary-models.asn\\\"\", fillcolor=\"#D5AEFC\"];\n  \"thresholds\" -> \"Run_GeneMark_Training\" [label=\"thr\"];\n  \"default2\" -> \"Run_GeneMark_Training\" [label=\"marked_annotation_name\"];\n  \"default2\" [label=\"\\\"marked-annotation.asn\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default3\" -> \"Run_GeneMark_Training\" [label=\"min_seq_len\"];\n  \"default3\" [label=\"200\", fillcolor=\"#D5AEFC\"];\n  \"default4\" -> \"Run_GeneMark_Training\" [label=\"nogenbank\"];\n  \"default4\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"inseq\" -> \"Run_GeneMark_Training\" [label=\"sequences\"];\n  \"asn_cache\" -> \"Run_GeneMark_Training\" [label=\"asn_cache\"];\n  \"uniColl_cache\" -> \"Run_GeneMark_Training\" [label=\"asn_cache\"];\n  \"default5\" -> \"Run_GeneMark_Training\" [label=\"tmp_dir_name\"];\n  \"default5\" [label=\"\\\"workdir\\\"\", fillcolor=\"#D5AEFC\"];\n  \"Get_ORFs\" -> \"Get_off_frame_ORFs\" [label=\"seq_entries\"];\n  \"Map_HMM_Hits\" -> \"Get_off_frame_ORFs\" [label=\"aligns\"];\n  \"Search_All_HMMs_I\" -> \"Map_HMM_Hits\" [label=\"hmm_hits\"];\n  \"default6\" -> \"Map_HMM_Hits\" [label=\"align_fmt\"];\n  \"default6\" [label=\"\\\"seq-align\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default7\" -> \"Map_HMM_Hits\" [label=\"expansion_ratio\"];\n  \"default7\" [label=\"0e+00\", fillcolor=\"#D5AEFC\"];\n  \"Get_ORFs\" -> \"Map_HMM_Hits\" [label=\"sequences\"];\n  \"default8\" -> \"Map_HMM_Hits\" [label=\"no_compart\"];\n  \"default8\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"default9\" -> \"Map_HMM_Hits\" [label=\"nogenbank\"];\n  \"default9\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"asn_cache\" -> \"Map_HMM_Hits\" [label=\"seq_cache\"];\n  \"asn_cache\" -> \"Map_HMM_Hits\" [label=\"asn_cache\"];\n  \"uniColl_cache\" -> \"Map_HMM_Hits\" [label=\"asn_cache\"];\n  \"uniColl_cache\" -> \"Map_HMM_Hits\" [label=\"unicoll_cache\"];\n  \"inseq\" -> \"Get_ORFs\" [label=\"input\"];\n  \"asn_cache\" -> \"Get_ORFs\" [label=\"asn_cache\"];\n  \"hmms_tab\" -> \"Search_All_HMMs_I\" [label=\"hmms_tab\"];\n  \"scatter_gather_nchunks\" -> \"Search_All_HMMs_I\" [label=\"scatter_gather_nchunks\"];\n  \"Extract_ORF_Proteins\" -> \"Search_All_HMMs_I\" [label=\"proteins\"];\n  \"Extract_ORF_Proteins\" -> \"Search_All_HMMs_I\" [label=\"lds2\"];\n  \"asn_cache\" -> \"Search_All_HMMs_I\" [label=\"asn_cache\"];\n  \"hmm_path\" -> \"Search_All_HMMs_I\" [label=\"hmm_path\"];\n  \"Extract_ORF_Proteins\" -> \"Search_All_HMMs_I\" [label=\"seqids\"];\n  \"Get_ORFs\" -> \"Extract_ORF_Proteins\" [label=\"input\"];\n  \"nogenbank\" -> \"Extract_ORF_Proteins\" [label=\"nogenbank\"];\n  \"asn_cache\" -> \"Resolve_Annotation_Conflicts\" [label=\"asn_cache\"];\n  \"Execute_CRISPRs_annots\" -> \"Resolve_Annotation_Conflicts\" [label=\"features\"];\n  \"Generate_16S_rRNA_Annotation_annotation\" -> \"Resolve_Annotation_Conflicts\" [label=\"features\"];\n  \"Generate_23S_rRNA_Annotation_annotation\" -> \"Resolve_Annotation_Conflicts\" [label=\"features\"];\n  \"Post_process_CMsearch_annotations_annots_5S\" -> \"Resolve_Annotation_Conflicts\" [label=\"features\"];\n  \"trna_annots\" -> \"Resolve_Annotation_Conflicts\" [label=\"features\"];\n  \"ncrna_annots\" -> \"Resolve_Annotation_Conflicts\" [label=\"features\"];\n  \"thresholds\" -> \"Resolve_Annotation_Conflicts\" [label=\"thr\"];\n  \"Resolve_Annotation_Conflicts\" -> \"protein_aligns\";\n  \"Extract_ORF_Proteins\" -> \"proteins\";\n  \"Map_HMM_Hits\" -> \"aligns\";\n  \"Search_All_HMMs_I\" -> \"hmm_hits\";\n  \"Get_ORFs\" -> \"outseqs\";\n  \"Extract_ORF_Proteins\" -> \"seqids\";\n  \"Extract_ORF_Proteins\" -> \"lds2\";\n  \"Run_GeneMark_Training\" -> \"out_hmm_params\";\n  \"Get_off_frame_ORFs\" -> \"prot_ids\";\n  \"Resolve_Annotation_Conflicts\" -> \"annotation\";\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_pass1.cwl","packed":false,"visualisationXdot":"/graph/xdot/github.com/ncbi/pgap/blob/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl","visualisationPng":"/graph/png/github.com/ncbi/pgap/blob/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl","visualisationSvg":"/graph/svg/github.com/ncbi/pgap/blob/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl","roBundle":"/robundle/github.com/ncbi/pgap/blob/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/bacterial_annot/wf_bacterial_annot_pass1.cwl","licenseName":"https://raw.githubusercontent.com/ncbi/pgap/7f857f7f2d7c080d27c775b67a6d6f7d94bce31f/LICENSE.md"}