{"@id":"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl","retrievedFrom":{"repoUrl":"https://github.com/ncbi/pgap.git","branch":"13b411d94a568e52797ba936732dc65f1a91bd41","path":"bacterial_annot/wf_bacterial_annot_pass4.cwl","packedId":null,"type":"GITHUB","rawUrl":"https://raw.githubusercontent.com/ncbi/pgap/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl","url":"https://github.com/ncbi/pgap/blob/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl"},"retrievedOn":"2026-08-27T03:58:52.296+00:00","lastCommit":"13b411d94a568e52797ba936732dc65f1a91bd41","label":"Bacterial Annotation, pass 4, blastp-based functional annotation (second pass)","inputs":{"lds2":{"type":"File"},"CDDdata":{"type":"Directory"},"CDDdata2":{"type":"Directory"},"proteins":{"type":"File"},"annotation":{"type":"File"},"thresholds":{"type":"File"},"naming_sqlite":{"type":"File"},"uniColl_cache":{"type":"Directory"},"blast_rules_db":{"type":"String"},"sequence_cache":{"type":"Directory"},"wp_assignments":{"type":"File"},"hmm_assignments":{"type":"File"},"naming_blast_db":{"type":"Directory"},"blast_rules_db_dir":{"type":"Directory"},"defline_cleanup_rules":{"type":"File"},"identification_db_dir":{"type":"Directory"},"scatter_gather_nchunks":{"type":"String"},"Extract_Model_Proteins_prot_ids":{"type":"File"},"Good_AntiFam_filtered_proteins_gilist":{"type":"File"}},"outputs":{"out_annotation":{"type":"File"}},"steps":{"Add_Names_to_Proteins":{"label":"add_prot_names_to_annot","run":"../progs/add_prot_names_to_annot.cwl","runType":"COMMANDLINETOOL","sources":{"Add_Names_to_Proteins":{"sourceIDs":["annotation"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Add_Names_to_Proteins/proteins":{"sourceIDs":["Assign_Clusters_to_Proteins"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Add_Names_to_Proteins/asn_cache":{"sourceIDs":["sequence_cache"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Add_Names_to_Proteins/nogenbank":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Add_Names_to_Proteins/unicoll_sqlite":{"sourceIDs":["naming_sqlite"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Add_Names_to_Proteins/defline_cleanup_rules":{"sourceIDs":["defline_cleanup_rules"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Add_Names_to_Proteins/submission_mode_genbank":{"defaultVal":"true"}}},"Bacterial_Annot_Filter":{"label":"bact_annot_filter","run":"../progs/bact_annot_filter.cwl","runType":"COMMANDLINETOOL","sources":{"Bacterial_Annot_Filter":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Bacterial_Annot_Filter/thr":{"sourceIDs":["thresholds"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Bacterial_Annot_Filter/input":{"sourceIDs":["Add_Names_to_Proteins"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Bacterial_Annot_Filter/asn_cache":{"sourceIDs":["sequence_cache"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Bacterial_Annot_Filter/max_overlap":{"defaultVal":"120"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Bacterial_Annot_Filter/long_model_limit":{"defaultVal":"1000000"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Bacterial_Annot_Filter/short_model_limit":{"defaultVal":"180"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Bacterial_Annot_Filter/abs_short_model_limit":{"defaultVal":"60"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Bacterial_Annot_Filter/max_unannotated_region":{"defaultVal":"5000"}}},"Find_best_protein_hits":{"label":"align_filter","run":"../progs/align_filter.cwl","runType":"COMMANDLINETOOL","sources":{"Find_best_protein_hits":{"defaultVal":"\\\"subject_coverage >= 10\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_best_protein_hits/ifmt":{"defaultVal":"\\\"seq-align-set\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_best_protein_hits/input":{"sourceIDs":["Find_Naming_Protein_Hits"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_best_protein_hits/asn_cache":{"sourceIDs":["sequence_cache"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_best_protein_hits/nogenbank":{"defaultVal":"true"}}},"Prepare_SPARCLBL_input":{"label":"prepare_sparclbl_input","run":"../progs/prepare_sparclbl_input.cwl","runType":"COMMANDLINETOOL","sources":{"Prepare_SPARCLBL_input":{"sourceIDs":["naming_sqlite"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Prepare_SPARCLBL_input/input":{"sourceIDs":["Extract_Model_Proteins_prot_ids"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Prepare_SPARCLBL_input/other_assignments":{"sourceIDs":["Assign_Clusters_to_Proteins"]}}},"Find_Naming_Protein_Hits":{"label":"blastp_wnode_naming","run":"../task_types/tt_blastp_wnode_naming.cwl","runType":"WORKFLOW","sources":{"Find_Naming_Protein_Hits":{"defaultVal":"\\\"score>0 && pct_identity_gapopen_only > 35\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/ids":{"sourceIDs":["Good_AntiFam_filtered_proteins_gilist"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/seg":{"defaultVal":"\\\"no\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/lds2":{"sourceIDs":["lds2"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/ofmt":{"defaultVal":"\\\"asn-binary\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/delay":{"defaultVal":"0"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/dbsize":{"defaultVal":"\\\"6000000000\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/evalue":{"defaultVal":"1e-01"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/blastdb":{"defaultVal":"\\\"blast_rules_db\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/compart":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/affinity":{"defaultVal":"\\\"subject\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/max_jobs":{"defaultVal":"1"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/no_merge":{"defaultVal":"false"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/proteins":{"sourceIDs":["proteins"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/asn_cache":{"sourceIDs":["sequence_cache"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/nogenbank":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/threshold":{"defaultVal":"21"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/word_size":{"defaultVal":"6"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/blastdb_dir":{"sourceIDs":["blast_rules_db_dir"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/top_by_score":{"defaultVal":"10"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/max_target_seqs":{"defaultVal":"50"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/comp_based_stats":{"defaultVal":"\\\"F\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/max_batch_length":{"defaultVal":"10000"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/allow_intersection":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Find_Naming_Protein_Hits/scatter_gather_nchunks":{"sourceIDs":["scatter_gather_nchunks"]}}},"Assign_Clusters_to_Proteins":{"label":"assign_cluster","run":"../progs/assign_cluster.cwl","runType":"COMMANDLINETOOL","sources":{"Assign_Clusters_to_Proteins":{"defaultVal":"\\\"blastdb\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/seg":{"defaultVal":"\\\"no\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/hfmt":{"defaultVal":"\\\"seq-align\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/hits":{"sourceIDs":["Assign_Clusters_to_Proteins_sort"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/lds2":{"sourceIDs":["lds2"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/task":{"defaultVal":"\\\"blastp\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/cutoff":{"defaultVal":"5e-01"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/margin":{"defaultVal":"5e-02"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/proteins":{"sourceIDs":["proteins"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/asn_cache":{"sourceIDs":["sequence_cache"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/nogenbank":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/threshold":{"defaultVal":"21"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/word_size":{"defaultVal":"6"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/namedb_dir":{"sourceIDs":["naming_blast_db"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/sure_cutoff":{"defaultVal":"1.5e-01"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/unicoll_sqlite":{"sourceIDs":["naming_sqlite"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins/comp_based_stats":{"defaultVal":"\\\"F\\\""}}},"Assign_Clusters_to_Proteins_sort":{"label":"align_sort","run":"../progs/align_sort.cwl","runType":"COMMANDLINETOOL","sources":{"Assign_Clusters_to_Proteins_sort":{"defaultVal":"true"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins_sort/k":{"defaultVal":"\\\"query,subject,-score,-num_ident,query_align_len,subject_align_len,query_start,subject_start\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins_sort/ifmt":{"defaultVal":"\\\"seq-align-set\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins_sort/input":{"sourceIDs":["Find_best_protein_hits"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_Clusters_to_Proteins_sort/limit_mem":{"defaultVal":"\\\"13G\\\""}}},"Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl":{"label":"sparclbl","run":"../progs/sparclbl.cwl","runType":"COMMANDLINETOOL","sources":{"Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl":{"defaultVal":"1"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl/b":{"sourceIDs":["CDDdata"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl/d":{"sourceIDs":["CDDdata2"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl/m":{"defaultVal":"20"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl/n":{"defaultVal":"500"},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl/p":{"sourceIDs":["Prepare_SPARCLBL_input"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl/s":{"sourceIDs":["Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta"]}}},"Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta":{"label":"asn2fasta","run":"../progs/asn2fasta.cwl","runType":"COMMANDLINETOOL","sources":{"Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta":{"sourceIDs":["Assign_SPARCL_Architecture_Names_to_Proteins_gp_fetch_sequences"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta/type":{"defaultVal":"\\\"seq-entry\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta/serial":{"defaultVal":"\\\"binary\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta/fasta_name":{"defaultVal":"\\\"proteins.fa\\\""},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta/prots_only":{"defaultVal":"true"}}},"Assign_SPARCL_Architecture_Names_to_Proteins_gp_fetch_sequences":{"label":"gp_fetch_sequences","run":"../progs/gp_fetch_sequences.cwl","runType":"COMMANDLINETOOL","sources":{"Assign_SPARCL_Architecture_Names_to_Proteins_gp_fetch_sequences":{"sourceIDs":["lds2"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_SPARCL_Architecture_Names_to_Proteins_gp_fetch_sequences/input":{"sourceIDs":["Prepare_SPARCLBL_input"]},"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl#Assign_SPARCL_Architecture_Names_to_Proteins_gp_fetch_sequences/proteins":{"sourceIDs":["proteins"]}}}},"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    \"blast_rules_db\" [fillcolor=\"#94DDF4\",label=\"blast_rules_db\"];\n    \"thresholds\" [fillcolor=\"#94DDF4\",label=\"thresholds\"];\n    \"naming_sqlite\" [fillcolor=\"#94DDF4\",label=\"naming_sqlite\"];\n    \"annotation\" [fillcolor=\"#94DDF4\",label=\"annotation\"];\n    \"lds2\" [fillcolor=\"#94DDF4\",label=\"lds2\"];\n    \"scatter_gather_nchunks\" [fillcolor=\"#94DDF4\",label=\"scatter_gather_nchunks\"];\n    \"CDDdata2\" [fillcolor=\"#94DDF4\",label=\"CDDdata2\"];\n    \"uniColl_cache\" [fillcolor=\"#94DDF4\",label=\"uniColl_cache\"];\n    \"sequence_cache\" [fillcolor=\"#94DDF4\",label=\"sequence_cache\"];\n    \"hmm_assignments\" [fillcolor=\"#94DDF4\",label=\"hmm_assignments\"];\n    \"CDDdata\" [fillcolor=\"#94DDF4\",label=\"CDDdata\"];\n    \"identification_db_dir\" [fillcolor=\"#94DDF4\",label=\"identification_db_dir\"];\n    \"blast_rules_db_dir\" [fillcolor=\"#94DDF4\",label=\"blast_rules_db_dir\"];\n    \"proteins\" [fillcolor=\"#94DDF4\",label=\"proteins\"];\n    \"Good_AntiFam_filtered_proteins_gilist\" [fillcolor=\"#94DDF4\",label=\"Good_AntiFam_filtered_proteins_gilist\"];\n    \"wp_assignments\" [fillcolor=\"#94DDF4\",label=\"wp_assignments\"];\n    \"naming_blast_db\" [fillcolor=\"#94DDF4\",label=\"naming_blast_db\"];\n    \"Extract_Model_Proteins_prot_ids\" [fillcolor=\"#94DDF4\",label=\"Extract_Model_Proteins_prot_ids\"];\n    \"defline_cleanup_rules\" [fillcolor=\"#94DDF4\",label=\"defline_cleanup_rules\"];\n  }\n  subgraph cluster_outputs {\n    rank = \"same\";\n    style = \"dashed\";\n    labelloc = \"b\";\n    label = \"Workflow Outputs\";\n    \"out_annotation\" [fillcolor=\"#94DDF4\",label=\"out_annotation\"];\n  }\n  \"Assign_Clusters_to_Proteins_sort\" [label=\"align_sort\"];\n  \"Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl\" [label=\"sparclbl\"];\n  \"Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta\" [label=\"asn2fasta\"];\n  \"Find_best_protein_hits\" [label=\"align_filter\"];\n  \"Bacterial_Annot_Filter\" [label=\"bact_annot_filter\"];\n  \"Add_Names_to_Proteins\" [label=\"add_prot_names_to_annot\"];\n  \"Find_Naming_Protein_Hits\" [label=\"blastp_wnode_naming\", fillcolor=\"#F3CEA1\"];\n  \"Assign_SPARCL_Architecture_Names_to_Proteins_gp_fetch_sequences\" [label=\"gp_fetch_sequences\"];\n  \"Assign_Clusters_to_Proteins\" [label=\"assign_cluster\"];\n  \"Prepare_SPARCLBL_input\" [label=\"prepare_sparclbl_input\"];\n  \"default1\" -> \"Assign_Clusters_to_Proteins_sort\" [label=\"nogenbank\"];\n  \"default1\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"default2\" -> \"Assign_Clusters_to_Proteins_sort\" [label=\"limit_mem\"];\n  \"default2\" [label=\"\\\"13G\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default3\" -> \"Assign_Clusters_to_Proteins_sort\" [label=\"k\"];\n  \"default3\" [label=\"\\\"query,subject,-score,-num_ident,query_align_len,subject_align_len,query_start,subject_start\\\"\", fillcolor=\"#D5AEFC\"];\n  \"Find_best_protein_hits\" -> \"Assign_Clusters_to_Proteins_sort\" [label=\"input\"];\n  \"default4\" -> \"Assign_Clusters_to_Proteins_sort\" [label=\"ifmt\"];\n  \"default4\" [label=\"\\\"seq-align-set\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default5\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl\" [label=\"x\"];\n  \"default5\" [label=\"1\", fillcolor=\"#D5AEFC\"];\n  \"CDDdata\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl\" [label=\"b\"];\n  \"Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl\" [label=\"s\"];\n  \"CDDdata2\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl\" [label=\"d\"];\n  \"default6\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl\" [label=\"m\"];\n  \"default6\" [label=\"20\", fillcolor=\"#D5AEFC\"];\n  \"Prepare_SPARCLBL_input\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl\" [label=\"p\"];\n  \"default7\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl\" [label=\"n\"];\n  \"default7\" [label=\"500\", fillcolor=\"#D5AEFC\"];\n  \"Assign_SPARCL_Architecture_Names_to_Proteins_gp_fetch_sequences\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta\" [label=\"i\"];\n  \"default8\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta\" [label=\"prots_only\"];\n  \"default8\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"default9\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta\" [label=\"fasta_name\"];\n  \"default9\" [label=\"\\\"proteins.fa\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default10\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta\" [label=\"type\"];\n  \"default10\" [label=\"\\\"seq-entry\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default11\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_asn2fasta\" [label=\"serial\"];\n  \"default11\" [label=\"\\\"binary\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default12\" -> \"Find_best_protein_hits\" [label=\"filter\"];\n  \"default12\" [label=\"\\\"subject_coverage >= 10\\\"\", fillcolor=\"#D5AEFC\"];\n  \"Find_Naming_Protein_Hits\" -> \"Find_best_protein_hits\" [label=\"input\"];\n  \"default13\" -> \"Find_best_protein_hits\" [label=\"nogenbank\"];\n  \"default13\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"default14\" -> \"Find_best_protein_hits\" [label=\"ifmt\"];\n  \"default14\" [label=\"\\\"seq-align-set\\\"\", fillcolor=\"#D5AEFC\"];\n  \"uniColl_cache\" -> \"Find_best_protein_hits\" [label=\"asn_cache\"];\n  \"sequence_cache\" -> \"Find_best_protein_hits\" [label=\"asn_cache\"];\n  \"default15\" -> \"Bacterial_Annot_Filter\" [label=\"nogebank\"];\n  \"default15\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"thresholds\" -> \"Bacterial_Annot_Filter\" [label=\"thr\"];\n  \"default16\" -> \"Bacterial_Annot_Filter\" [label=\"max_overlap\"];\n  \"default16\" [label=\"120\", fillcolor=\"#D5AEFC\"];\n  \"default17\" -> \"Bacterial_Annot_Filter\" [label=\"long_model_limit\"];\n  \"default17\" [label=\"1000000\", fillcolor=\"#D5AEFC\"];\n  \"default18\" -> \"Bacterial_Annot_Filter\" [label=\"max_unannotated_region\"];\n  \"default18\" [label=\"5000\", fillcolor=\"#D5AEFC\"];\n  \"Add_Names_to_Proteins\" -> \"Bacterial_Annot_Filter\" [label=\"input\"];\n  \"default19\" -> \"Bacterial_Annot_Filter\" [label=\"abs_short_model_limit\"];\n  \"default19\" [label=\"60\", fillcolor=\"#D5AEFC\"];\n  \"default20\" -> \"Bacterial_Annot_Filter\" [label=\"short_model_limit\"];\n  \"default20\" [label=\"180\", fillcolor=\"#D5AEFC\"];\n  \"sequence_cache\" -> \"Bacterial_Annot_Filter\" [label=\"asn_cache\"];\n  \"annotation\" -> \"Add_Names_to_Proteins\" [label=\"input\"];\n  \"default21\" -> \"Add_Names_to_Proteins\" [label=\"nogenbank\"];\n  \"default21\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"naming_sqlite\" -> \"Add_Names_to_Proteins\" [label=\"unicoll_sqlite\"];\n  \"hmm_assignments\" -> \"Add_Names_to_Proteins\" [label=\"proteins\"];\n  \"Assign_SPARCL_Architecture_Names_to_Proteins_sparclbl\" -> \"Add_Names_to_Proteins\" [label=\"proteins\"];\n  \"wp_assignments\" -> \"Add_Names_to_Proteins\" [label=\"proteins\"];\n  \"Assign_Clusters_to_Proteins\" -> \"Add_Names_to_Proteins\" [label=\"proteins\"];\n  \"sequence_cache\" -> \"Add_Names_to_Proteins\" [label=\"asn_cache\"];\n  \"default22\" -> \"Add_Names_to_Proteins\" [label=\"submission_mode_genbank\"];\n  \"default22\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"defline_cleanup_rules\" -> \"Add_Names_to_Proteins\" [label=\"defline_cleanup_rules\"];\n  \"default23\" -> \"Find_Naming_Protein_Hits\" [label=\"align_filter\"];\n  \"default23\" [label=\"\\\"score>0 && pct_identity_gapopen_only > 35\\\"\", fillcolor=\"#D5AEFC\"];\n  \"lds2\" -> \"Find_Naming_Protein_Hits\" [label=\"lds2\"];\n  \"scatter_gather_nchunks\" -> \"Find_Naming_Protein_Hits\" [label=\"scatter_gather_nchunks\"];\n  \"default24\" -> \"Find_Naming_Protein_Hits\" [label=\"max_target_seqs\"];\n  \"default24\" [label=\"50\", fillcolor=\"#D5AEFC\"];\n  \"default25\" -> \"Find_Naming_Protein_Hits\" [label=\"seg\"];\n  \"default25\" [label=\"\\\"no\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default26\" -> \"Find_Naming_Protein_Hits\" [label=\"dbsize\"];\n  \"default26\" [label=\"\\\"6000000000\\\"\", fillcolor=\"#D5AEFC\"];\n  \"identification_db_dir\" -> \"Find_Naming_Protein_Hits\" [label=\"blastdb_dir\"];\n  \"blast_rules_db_dir\" -> \"Find_Naming_Protein_Hits\" [label=\"blastdb_dir\"];\n  \"default27\" -> \"Find_Naming_Protein_Hits\" [label=\"evalue\"];\n  \"default27\" [label=\"1e-01\", fillcolor=\"#D5AEFC\"];\n  \"default28\" -> \"Find_Naming_Protein_Hits\" [label=\"no_merge\"];\n  \"default28\" [label=\"false\", fillcolor=\"#D5AEFC\"];\n  \"default29\" -> \"Find_Naming_Protein_Hits\" [label=\"delay\"];\n  \"default29\" [label=\"0\", fillcolor=\"#D5AEFC\"];\n  \"default30\" -> \"Find_Naming_Protein_Hits\" [label=\"top_by_score\"];\n  \"default30\" [label=\"10\", fillcolor=\"#D5AEFC\"];\n  \"default31\" -> \"Find_Naming_Protein_Hits\" [label=\"max_jobs\"];\n  \"default31\" [label=\"1\", fillcolor=\"#D5AEFC\"];\n  \"proteins\" -> \"Find_Naming_Protein_Hits\" [label=\"proteins\"];\n  \"default32\" -> \"Find_Naming_Protein_Hits\" [label=\"threshold\"];\n  \"default32\" [label=\"21\", fillcolor=\"#D5AEFC\"];\n  \"default33\" -> \"Find_Naming_Protein_Hits\" [label=\"blastdb\"];\n  \"default33\" [label=\"\\\"blastdb\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default34\" -> \"Find_Naming_Protein_Hits\" [label=\"blastdb\"];\n  \"default34\" [label=\"\\\"blast_rules_db\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default35\" -> \"Find_Naming_Protein_Hits\" [label=\"comp_based_stats\"];\n  \"default35\" [label=\"\\\"F\\\"\", fillcolor=\"#D5AEFC\"];\n  \"Good_AntiFam_filtered_proteins_gilist\" -> \"Find_Naming_Protein_Hits\" [label=\"ids\"];\n  \"default36\" -> \"Find_Naming_Protein_Hits\" [label=\"max_batch_length\"];\n  \"default36\" [label=\"10000\", fillcolor=\"#D5AEFC\"];\n  \"default37\" -> \"Find_Naming_Protein_Hits\" [label=\"allow_intersection\"];\n  \"default37\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"uniColl_cache\" -> \"Find_Naming_Protein_Hits\" [label=\"asn_cache\"];\n  \"sequence_cache\" -> \"Find_Naming_Protein_Hits\" [label=\"asn_cache\"];\n  \"default38\" -> \"Find_Naming_Protein_Hits\" [label=\"word_size\"];\n  \"default38\" [label=\"6\", fillcolor=\"#D5AEFC\"];\n  \"default39\" -> \"Find_Naming_Protein_Hits\" [label=\"nogenbank\"];\n  \"default39\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"default40\" -> \"Find_Naming_Protein_Hits\" [label=\"affinity\"];\n  \"default40\" [label=\"\\\"subject\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default41\" -> \"Find_Naming_Protein_Hits\" [label=\"ofmt\"];\n  \"default41\" [label=\"\\\"asn-binary\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default42\" -> \"Find_Naming_Protein_Hits\" [label=\"compart\"];\n  \"default42\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"lds2\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_gp_fetch_sequences\" [label=\"lds2\"];\n  \"Prepare_SPARCLBL_input\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_gp_fetch_sequences\" [label=\"input\"];\n  \"proteins\" -> \"Assign_SPARCL_Architecture_Names_to_Proteins_gp_fetch_sequences\" [label=\"proteins\"];\n  \"default43\" -> \"Assign_Clusters_to_Proteins\" [label=\"namedb\"];\n  \"default43\" [label=\"\\\"blastdb\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default44\" -> \"Assign_Clusters_to_Proteins\" [label=\"seg\"];\n  \"default44\" [label=\"\\\"no\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default45\" -> \"Assign_Clusters_to_Proteins\" [label=\"cutoff\"];\n  \"default45\" [label=\"5e-01\", fillcolor=\"#D5AEFC\"];\n  \"uniColl_cache\" -> \"Assign_Clusters_to_Proteins\" [label=\"asn_cache\"];\n  \"sequence_cache\" -> \"Assign_Clusters_to_Proteins\" [label=\"asn_cache\"];\n  \"naming_sqlite\" -> \"Assign_Clusters_to_Proteins\" [label=\"unicoll_sqlite\"];\n  \"default46\" -> \"Assign_Clusters_to_Proteins\" [label=\"comp_based_stats\"];\n  \"default46\" [label=\"\\\"F\\\"\", fillcolor=\"#D5AEFC\"];\n  \"Assign_Clusters_to_Proteins_sort\" -> \"Assign_Clusters_to_Proteins\" [label=\"hits\"];\n  \"default47\" -> \"Assign_Clusters_to_Proteins\" [label=\"threshold\"];\n  \"default47\" [label=\"21\", fillcolor=\"#D5AEFC\"];\n  \"default48\" -> \"Assign_Clusters_to_Proteins\" [label=\"sure_cutoff\"];\n  \"default48\" [label=\"1.5e-01\", fillcolor=\"#D5AEFC\"];\n  \"lds2\" -> \"Assign_Clusters_to_Proteins\" [label=\"lds2\"];\n  \"default49\" -> \"Assign_Clusters_to_Proteins\" [label=\"margin\"];\n  \"default49\" [label=\"5e-02\", fillcolor=\"#D5AEFC\"];\n  \"proteins\" -> \"Assign_Clusters_to_Proteins\" [label=\"proteins\"];\n  \"default50\" -> \"Assign_Clusters_to_Proteins\" [label=\"nogenbank\"];\n  \"default50\" [label=\"true\", fillcolor=\"#D5AEFC\"];\n  \"naming_blast_db\" -> \"Assign_Clusters_to_Proteins\" [label=\"namedb_dir\"];\n  \"default51\" -> \"Assign_Clusters_to_Proteins\" [label=\"word_size\"];\n  \"default51\" [label=\"6\", fillcolor=\"#D5AEFC\"];\n  \"default52\" -> \"Assign_Clusters_to_Proteins\" [label=\"hfmt\"];\n  \"default52\" [label=\"\\\"seq-align\\\"\", fillcolor=\"#D5AEFC\"];\n  \"default53\" -> \"Assign_Clusters_to_Proteins\" [label=\"task\"];\n  \"default53\" [label=\"\\\"blastp\\\"\", fillcolor=\"#D5AEFC\"];\n  \"naming_sqlite\" -> \"Prepare_SPARCLBL_input\" [label=\"unicoll_sqlite\"];\n  \"hmm_assignments\" -> \"Prepare_SPARCLBL_input\" [label=\"other_assignments\"];\n  \"wp_assignments\" -> \"Prepare_SPARCLBL_input\" [label=\"other_assignments\"];\n  \"Assign_Clusters_to_Proteins\" -> \"Prepare_SPARCLBL_input\" [label=\"other_assignments\"];\n  \"Extract_Model_Proteins_prot_ids\" -> \"Prepare_SPARCLBL_input\" [label=\"input\"];\n  \"Bacterial_Annot_Filter\" -> \"out_annotation\";\n}\n","licenseLink":"https://raw.githubusercontent.com/ncbi/pgap/13b411d94a568e52797ba936732dc65f1a91bd41/LICENSE.md","permalink":"https://w3id.org/cwl/view/git/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl","packed":false,"visualisationXdot":"/graph/xdot/github.com/ncbi/pgap/blob/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl","visualisationPng":"/graph/png/github.com/ncbi/pgap/blob/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl","visualisationSvg":"/graph/svg/github.com/ncbi/pgap/blob/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl","roBundle":"/robundle/github.com/ncbi/pgap/blob/13b411d94a568e52797ba936732dc65f1a91bd41/bacterial_annot/wf_bacterial_annot_pass4.cwl","licenseName":"https://raw.githubusercontent.com/ncbi/pgap/13b411d94a568e52797ba936732dc65f1a91bd41/LICENSE.md"}