{"@id":"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl","retrievedFrom":{"repoUrl":"https://github.com/ncbi/pgap.git","branch":"01a5c0a8834846ce04fed190eec7d1cc39a3df48","path":"protein_alignment/wf_protein_alignment_miniprot.cwl","packedId":null,"type":"GITHUB","rawUrl":"https://raw.githubusercontent.com/ncbi/pgap/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl","url":"https://github.com/ncbi/pgap/blob/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl"},"retrievedOn":"2026-09-06T11:05:07.172+00:00","lastCommit":"01a5c0a8834846ce04fed190eec7d1cc39a3df48","label":"Align reference proteins plane complete workflow, with miniprot","inputs":{"go":{"type":"Boolean[]"},"taxid":{"type":"Integer"},"taxon_db":{"type":"File"},"all_prots":{"type":"File"},"asn_cache":{"type":"Directory"},"blastdb_dir":{"type":"Directory"},"genomic_ids":{"type":"File"},"compartments":{"type":"File"},"uniColl_asn_cache":{"type":"Directory"}},"outputs":{"align":{"type":"File"}},"steps":{"Compute_Gencode":{"label":"compute_gencode","run":"../progs/compute_gencode.cwl","runType":"COMMANDLINETOOL","sources":{"Compute_Gencode":{"sourceIDs":["taxon_db"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Compute_Gencode/taxid":{"sourceIDs":["taxid"]}}},"Get_Genomic_FASTA":{"label":"cross_origin_fasta","run":"../progs/cross_origin_fasta.cwl","runType":"COMMANDLINETOOL","sources":{"Get_Genomic_FASTA":{"sourceIDs":["genomic_ids"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Get_Genomic_FASTA/asn_cache":{"sourceIDs":["asn_cache"]}}},"Get_Protein_FASTA":{"label":"getfasta","run":"../progs/getfasta.cwl","runType":"COMMANDLINETOOL","sources":{"Get_Protein_FASTA":{"sourceIDs":["uniColl_asn_cache"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Get_Protein_FASTA/gilist":{"sourceIDs":["Extract_Proteins_From_Compartments"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Get_Protein_FASTA/asn_cache":{"sourceIDs":["asn_cache"]}}},"Run_Miniprot_fast":{"label":"miniprot","run":"../progs/miniprot.cwl","runType":"COMMANDLINETOOL","sources":{"Run_Miniprot_fast":{"sourceIDs":["Get_Protein_FASTA"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Run_Miniprot_fast/T":{"sourceIDs":["Compute_Gencode_int"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Run_Miniprot_fast/genome":{"sourceIDs":["Get_Genomic_FASTA"]}}},"Run_Miniprot_slow":{"label":"miniprot","run":"../progs/miniprot.cwl","runType":"COMMANDLINETOOL","sources":{"Run_Miniprot_slow":{"sourceIDs":["Compute_Gencode_int"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Run_Miniprot_slow/genome":{"sourceIDs":["Get_Genomic_FASTA"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Run_Miniprot_slow/proteins":{"sourceIDs":["Get_Protein_FASTA"]}}},"Compute_Gencode_int":{"run":"../progs/file2int.cwl","runType":"EXPRESSIONTOOL","sources":{"Compute_Gencode_int":{"sourceIDs":["Compute_Gencode"]}}},"Align_Short_Proteins":{"label":"Seed Protein Alignments","run":"wf_seed_seqids.cwl","runType":"WORKFLOW","sources":{"Align_Short_Proteins":{"sourceIDs":["Compute_Gencode_int"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Align_Short_Proteins/seqids":{"sourceIDs":["Filter_Short_Sequences"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Align_Short_Proteins/asn_cache":{"sourceIDs":["asn_cache"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Align_Short_Proteins/blastdb_dir":{"sourceIDs":["blastdb_dir"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Align_Short_Proteins/uniColl_asn_cache":{"sourceIDs":["uniColl_asn_cache"]}}},"Convert_PAF_Alignments":{"label":"paf2asn","run":"../progs/paf2asn.cwl","runType":"COMMANDLINETOOL","sources":{"Convert_PAF_Alignments":{"sourceIDs":["uniColl_asn_cache"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Convert_PAF_Alignments/paf":{"sourceIDs":["Run_Miniprot_slow"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Convert_PAF_Alignments/asn_cache":{"sourceIDs":["asn_cache"]}}},"Filter_Short_Sequences":{"label":"seq_filter","run":"../progs/seq_filter.cwl","runType":"COMMANDLINETOOL","sources":{"Filter_Short_Sequences":{"sourceIDs":["uniColl_asn_cache"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Filter_Short_Sequences/seqids":{"sourceIDs":["Extract_Proteins_From_Compartments"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Filter_Short_Sequences/asn_cache":{"sourceIDs":["asn_cache"]}}},"Sort_Protein_Alignments":{"label":"align_sort","run":"../progs/align_sort.cwl","runType":"COMMANDLINETOOL","sources":{"Sort_Protein_Alignments":{"sourceIDs":["Filter_Protein_Compartments"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Sort_Protein_Alignments/asn_cache":{"sourceIDs":["asn_cache"]}}},"Filter_Protein_Alignments":{"label":"Filter Protein Alignments","run":"align_filter.cwl","runType":"COMMANDLINETOOL","sources":{"Filter_Protein_Alignments":{"sourceIDs":["Sort_Protein_Alignments"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Filter_Protein_Alignments/asn_cache":{"sourceIDs":["asn_cache"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Filter_Protein_Alignments/uniColl_asn_cache":{"sourceIDs":["uniColl_asn_cache"]}}},"Filter_Protein_Compartments":{"label":"align_filter","run":"../progs/align_filter.cwl","runType":"COMMANDLINETOOL","sources":{"Filter_Protein_Compartments":{"sourceIDs":["Remap_cross_origin_alignments"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Filter_Protein_Compartments/input":{"sourceIDs":["Filter_Full_Coverage_Alignments"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Filter_Protein_Compartments/asn_cache":{"sourceIDs":["asn_cache"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Filter_Protein_Compartments/compartments":{"sourceIDs":["compartments"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Filter_Protein_Compartments/subject_allowlist":{"sourceIDs":["all_prots"]}}},"Remap_cross_origin_alignments":{"label":"cross_origin_fasta","run":"../progs/cross_origin_fasta.cwl","runType":"COMMANDLINETOOL","sources":{"Remap_cross_origin_alignments":{"sourceIDs":["Get_Genomic_FASTA"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Remap_cross_origin_alignments/align":{"sourceIDs":["Convert_PAF_Alignments"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Remap_cross_origin_alignments/asn_cache":{"sourceIDs":["asn_cache"]}}},"Sort_Short_Protein_Alignments":{"label":"align_sort_ma","run":"../progs/align_sort_ma.cwl","runType":"COMMANDLINETOOL","sources":{"Sort_Short_Protein_Alignments":{"sourceIDs":["Align_Short_Proteins"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Sort_Short_Protein_Alignments/asn_cache":{"sourceIDs":["asn_cache"]}}},"Filter_Full_Coverage_Alignments":{"label":"Filter Full-Coverage Alignments","run":"bacterial_protalign_filter.cwl","runType":"COMMANDLINETOOL","sources":{"Filter_Full_Coverage_Alignments":{"sourceIDs":["asn_cache"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Filter_Full_Coverage_Alignments/sorted_seeds":{"sourceIDs":["Sort_Short_Protein_Alignments"]},"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl#Filter_Full_Coverage_Alignments/uniColl_asn_cache":{"sourceIDs":["uniColl_asn_cache"]}}},"Extract_Proteins_From_Compartments":{"label":"compart_annot2all","run":"../progs/compart_annot2all.cwl","runType":"COMMANDLINETOOL","sources":{"Extract_Proteins_From_Compartments":{"sourceIDs":["compartments"]}}}},"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    \"compartments\" [fillcolor=\"#94DDF4\",label=\"compartments\"];\n    \"taxon_db\" [fillcolor=\"#94DDF4\",label=\"taxon_db\"];\n    \"all_prots\" [fillcolor=\"#94DDF4\",label=\"all_prots\"];\n    \"genomic_ids\" [fillcolor=\"#94DDF4\",label=\"genomic_ids\"];\n    \"go\" [fillcolor=\"#94DDF4\",label=\"go\"];\n    \"blastdb_dir\" [fillcolor=\"#94DDF4\",label=\"blastdb_dir\"];\n    \"asn_cache\" [fillcolor=\"#94DDF4\",label=\"asn_cache\"];\n    \"taxid\" [fillcolor=\"#94DDF4\",label=\"taxid\"];\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  }\n  \"Get_Genomic_FASTA\" [label=\"cross_origin_fasta\"];\n  \"Extract_Proteins_From_Compartments\" [label=\"compart_annot2all\"];\n  \"Filter_Full_Coverage_Alignments\" [label=\"Filter Full-Coverage Alignments\"];\n  \"Align_Short_Proteins\" [label=\"Seed Protein Alignments\", fillcolor=\"#F3CEA1\"];\n  \"Filter_Protein_Alignments\" [label=\"Filter Protein Alignments\"];\n  \"Run_Miniprot_slow\" [label=\"miniprot\"];\n  \"Remap_cross_origin_alignments\" [label=\"cross_origin_fasta\"];\n  \"Convert_PAF_Alignments\" [label=\"paf2asn\"];\n  \"Filter_Protein_Compartments\" [label=\"align_filter\"];\n  \"Compute_Gencode_int\" [label=\"Compute_Gencode_int\"];\n  \"Sort_Short_Protein_Alignments\" [label=\"align_sort_ma\"];\n  \"Sort_Protein_Alignments\" [label=\"align_sort\"];\n  \"Filter_Short_Sequences\" [label=\"seq_filter\"];\n  \"Run_Miniprot_fast\" [label=\"miniprot\"];\n  \"Compute_Gencode\" [label=\"compute_gencode\"];\n  \"Get_Protein_FASTA\" [label=\"getfasta\"];\n  \"genomic_ids\" -> \"Get_Genomic_FASTA\" [label=\"gilist\"];\n  \"asn_cache\" -> \"Get_Genomic_FASTA\" [label=\"asn_cache\"];\n  \"compartments\" -> \"Extract_Proteins_From_Compartments\" [label=\"aligns\"];\n  \"asn_cache\" -> \"Filter_Full_Coverage_Alignments\" [label=\"asn_cache\"];\n  \"Sort_Short_Protein_Alignments\" -> \"Filter_Full_Coverage_Alignments\" [label=\"sorted_seeds\"];\n  \"uniColl_asn_cache\" -> \"Filter_Full_Coverage_Alignments\" [label=\"uniColl_asn_cache\"];\n  \"Compute_Gencode_int\" -> \"Align_Short_Proteins\" [label=\"db_gencode\"];\n  \"blastdb_dir\" -> \"Align_Short_Proteins\" [label=\"blastdb_dir\"];\n  \"uniColl_asn_cache\" -> \"Align_Short_Proteins\" [label=\"uniColl_asn_cache\"];\n  \"Filter_Short_Sequences\" -> \"Align_Short_Proteins\" [label=\"seqids\"];\n  \"asn_cache\" -> \"Align_Short_Proteins\" [label=\"asn_cache\"];\n  \"Sort_Protein_Alignments\" -> \"Filter_Protein_Alignments\" [label=\"file_in\"];\n  \"uniColl_asn_cache\" -> \"Filter_Protein_Alignments\" [label=\"uniColl_asn_cache\"];\n  \"asn_cache\" -> \"Filter_Protein_Alignments\" [label=\"asn_cache\"];\n  \"Compute_Gencode_int\" -> \"Run_Miniprot_slow\" [label=\"T\"];\n  \"Get_Genomic_FASTA\" -> \"Run_Miniprot_slow\" [label=\"genome\"];\n  \"Get_Protein_FASTA\" -> \"Run_Miniprot_slow\" [label=\"proteins\"];\n  \"Get_Genomic_FASTA\" -> \"Remap_cross_origin_alignments\" [label=\"delta_seqs_input\"];\n  \"uniColl_asn_cache\" -> \"Remap_cross_origin_alignments\" [label=\"asn_cache\"];\n  \"asn_cache\" -> \"Remap_cross_origin_alignments\" [label=\"asn_cache\"];\n  \"Convert_PAF_Alignments\" -> \"Remap_cross_origin_alignments\" [label=\"align\"];\n  \"uniColl_asn_cache\" -> \"Convert_PAF_Alignments\" [label=\"asn_cache\"];\n  \"asn_cache\" -> \"Convert_PAF_Alignments\" [label=\"asn_cache\"];\n  \"Run_Miniprot_fast\" -> \"Convert_PAF_Alignments\" [label=\"paf\"];\n  \"Run_Miniprot_slow\" -> \"Convert_PAF_Alignments\" [label=\"paf\"];\n  \"Remap_cross_origin_alignments\" -> \"Filter_Protein_Compartments\" [label=\"input\"];\n  \"Filter_Full_Coverage_Alignments\" -> \"Filter_Protein_Compartments\" [label=\"input\"];\n  \"asn_cache\" -> \"Filter_Protein_Compartments\" [label=\"asn_cache\"];\n  \"all_prots\" -> \"Filter_Protein_Compartments\" [label=\"subject_allowlist\"];\n  \"compartments\" -> \"Filter_Protein_Compartments\" [label=\"compartments\"];\n  \"Compute_Gencode\" -> \"Compute_Gencode_int\" [label=\"input\"];\n  \"Align_Short_Proteins\" -> \"Sort_Short_Protein_Alignments\" [label=\"input\"];\n  \"uniColl_asn_cache\" -> \"Sort_Short_Protein_Alignments\" [label=\"asn_cache\"];\n  \"asn_cache\" -> \"Sort_Short_Protein_Alignments\" [label=\"asn_cache\"];\n  \"Filter_Protein_Compartments\" -> \"Sort_Protein_Alignments\" [label=\"input\"];\n  \"asn_cache\" -> \"Sort_Protein_Alignments\" [label=\"asn_cache\"];\n  \"uniColl_asn_cache\" -> \"Filter_Short_Sequences\" [label=\"asn_cache\"];\n  \"asn_cache\" -> \"Filter_Short_Sequences\" [label=\"asn_cache\"];\n  \"all_prots\" -> \"Filter_Short_Sequences\" [label=\"seqids\"];\n  \"Extract_Proteins_From_Compartments\" -> \"Filter_Short_Sequences\" [label=\"seqids\"];\n  \"Get_Protein_FASTA\" -> \"Run_Miniprot_fast\" [label=\"proteins\"];\n  \"Compute_Gencode_int\" -> \"Run_Miniprot_fast\" [label=\"T\"];\n  \"Get_Genomic_FASTA\" -> \"Run_Miniprot_fast\" [label=\"genome\"];\n  \"taxon_db\" -> \"Compute_Gencode\" [label=\"taxon_db\"];\n  \"taxid\" -> \"Compute_Gencode\" [label=\"taxid\"];\n  \"uniColl_asn_cache\" -> \"Get_Protein_FASTA\" [label=\"asn_cache\"];\n  \"asn_cache\" -> \"Get_Protein_FASTA\" [label=\"asn_cache\"];\n  \"all_prots\" -> \"Get_Protein_FASTA\" [label=\"gilist\"];\n  \"Extract_Proteins_From_Compartments\" -> \"Get_Protein_FASTA\" [label=\"gilist\"];\n  \"Filter_Protein_Alignments\" -> \"align\";\n}\n","licenseLink":"https://raw.githubusercontent.com/ncbi/pgap/01a5c0a8834846ce04fed190eec7d1cc39a3df48/LICENSE.md","permalink":"https://w3id.org/cwl/view/git/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl","packed":false,"visualisationXdot":"/graph/xdot/github.com/ncbi/pgap/blob/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl","visualisationPng":"/graph/png/github.com/ncbi/pgap/blob/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl","visualisationSvg":"/graph/svg/github.com/ncbi/pgap/blob/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl","roBundle":"/robundle/github.com/ncbi/pgap/blob/01a5c0a8834846ce04fed190eec7d1cc39a3df48/protein_alignment/wf_protein_alignment_miniprot.cwl","licenseName":"https://raw.githubusercontent.com/ncbi/pgap/01a5c0a8834846ce04fed190eec7d1cc39a3df48/LICENSE.md"}