@prefix LoadListingRequirement: . @prefix CommandLineBinding: . @prefix xsd: . @prefix DockerRequirement: . @prefix Workflow: . @prefix sld: . @prefix rdfs: . @prefix CommandOutputBinding: . @prefix ns1: . @prefix NetworkAccess: . @prefix rdf: . @prefix ScatterMethod: . @prefix cwl: . cwl:outputSource ; sld:type cwl:File . cwl:source ; cwl:valueFrom "$(self).superblat.sims.sort" . cwl:outputSource ; sld:type cwl:File . rdfs:comment "Output expanded sim file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outExpandName)" ] ; sld:type cwl:File . cwl:outputSource ; sld:type cwl:File . cwl:source . a cwl:CommandLineTool ; rdfs:comment "create abundance profile from expanded annotated sims files\nmd5: sims_abundance.py -t md5 -i -o --coverage --cluster --md5index \nlca: sims_abundance.py -t lca -i -o --coverage --cluster \nsource: sims_abundance.py -t source -i -o --coverage --cluster \n" ; rdfs:label "abundance profile" ; ns1:original_cwlVersion "v1.2" ; cwl:baseCommand ( "sims_abundance.py" ) ; cwl:baseCommand ( "sims_abundance.py" ) ; cwl:baseCommand ( "sims_abundance.py" ) ; cwl:cwlVersion ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:inputs , , , , , ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "sims_abundance.error" ; cwl:stdout "sims_abundance.log" . cwl:outputBinding [ CommandOutputBinding:glob "seqUtil.error" ] ; cwl:outputBinding [ CommandOutputBinding:glob "seqUtil.error" ] ; cwl:outputBinding [ CommandOutputBinding:glob "seqUtil.error" ] ; sld:type cwl:File . cwl:in , ; cwl:out ; cwl:run . cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.lenBin)" ] ; sld:type cwl:File , sld:null . cwl:outputBinding [ CommandOutputBinding:outputEval "$(inputs.length)" ] ; sld:type xsd:int . sld:type [ sld:items cwl:File ; sld:type sld:array ] . cwl:source . cwl:outputSource ; sld:type cwl:File . a cwl:Workflow ; rdfs:comment "Remove and trim low quality reads from fastq files. \nReturn fasta files with reads passed and reads removed.\n" ; rdfs:label "Preprocess fastq" ; ns1:original_cwlVersion "v1.0" ; Workflow:steps , , , ; cwl:cwlVersion ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , , , , ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:requirements [ a cwl:MultipleInputFeatureRequirement ] ; cwl:requirements [ a cwl:StepInputExpressionRequirement ] ; cwl:requirements [ a cwl:ScatterFeatureRequirement ] . cwl:source ; cwl:valueFrom "${ return [self]; }" . cwl:source . cwl:source . a cwl:CommandLineTool ; rdfs:comment "append feature sequence to each hit line of similarity file\n>add_seq2sims --seq_file --in_sim \n" ; rdfs:label "uncluster sims" ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( [ CommandLineBinding:prefix "--verbose" ] ) ; cwl:baseCommand ( "add_seq2sims" ) ; cwl:cwlVersion ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:inputs , , ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "add_seq2sims.error" ; cwl:stdout "add_seq2sims.log" . a cwl:Workflow ; rdfs:comment "create sorted / filtered similarity file with feature sequences, and index by md5" ; rdfs:label "index sim seq" ; ns1:original_cwlVersion "v1.0" ; Workflow:steps , , , , , , ; cwl:cwlVersion ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:inputs , , , ; cwl:outputs , ; cwl:requirements [ a cwl:MultipleInputFeatureRequirement ] ; cwl:requirements [ a cwl:StepInputExpressionRequirement ] ; cwl:requirements [ a cwl:ScatterFeatureRequirement ] ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] . cwl:source . cwl:outputBinding [ CommandOutputBinding:glob "sims_abundance.error" ] ; cwl:outputBinding [ CommandOutputBinding:glob "sims_abundance.error" ] ; cwl:outputBinding [ CommandOutputBinding:glob "sims_abundance.error" ] ; sld:type cwl:File . rdfs:comment "Input sequence file" ; cwl:inputBinding [ CommandLineBinding:prefix "--input" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--input" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--input" ] ; sld:type cwl:File . rdfs:comment "output prefix = ${output_prefix}.seq.bins, ${output_prefix}.seq.stats" ; cwl:inputBinding [ CommandLineBinding:prefix "-out_prefix" ] ; sld:type xsd:string . cwl:in , , , ; cwl:out , , ; cwl:run . rdfs:comment "Output .faa (protein) file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName).faa" ] ; sld:type cwl:File . cwl:source . cwl:outputSource ; sld:type cwl:File . a cwl:CommandLineTool ; rdfs:comment "multi-threaded fast sequence search command line tool, protein only\n>superblat -fastMap -prot -out blast8 \n" ; rdfs:label "superBLAT" ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( "-prot" "-out=blast8" ) ; cwl:baseCommand ( "superblat" ) ; cwl:cwlVersion ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:inputs , , , ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "superblat.error" ; cwl:stdout "superblat.log" . rdfs:comment "The sorted file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; sld:type cwl:File . cwl:source ; cwl:valueFrom "$(self).feature.seqs" . sld:type xsd:string . rdfs:comment "Length of kmer to use, eg. 6 or 15" ; cwl:default 6 ; cwl:inputBinding [ CommandLineBinding:prefix "--length" ] ; sld:type xsd:int , sld:null . cwl:outputSource ; sld:type cwl:File . cwl:source ; cwl:valueFrom "$(self).075" . cwl:source . rdfs:comment "Input similarity blast-m8 file" ; cwl:inputBinding [ CommandLineBinding:prefix "--in_file" ] ; sld:type cwl:File . cwl:default 30 ; sld:type xsd:int . cwl:valueFrom "$([\"1,1\"])" . cwl:source . cwl:in , , , , , ; cwl:out ; cwl:run . rdfs:comment "Optional input file, md5,seek,length" ; cwl:inputBinding [ CommandLineBinding:prefix "--md5_index" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--md5_index" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--md5_index" ] ; sld:type cwl:File , sld:null . cwl:outputSource ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "seqUtil.log" ] ; cwl:outputBinding [ CommandOutputBinding:glob "seqUtil.log" ] ; cwl:outputBinding [ CommandOutputBinding:glob "seqUtil.log" ] ; sld:type cwl:File . rdfs:comment "1 if the sequence file has complete genomic sequences\n0 if the sequence file has short sequence reads\ndefault is 0\n" ; cwl:default 0 ; cwl:inputBinding [ CommandLineBinding:prefix "--complete" ] ; sld:type xsd:int , sld:null . cwl:source ; cwl:valueFrom "$(self).700.annotation.md5.abundance" . rdfs:comment "Output fasta format file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; sld:type cwl:File . rdfs:comment "List of files to concatenate" ; cwl:inputBinding [ CommandLineBinding:position 1 ] ; cwl:inputBinding [ CommandLineBinding:position 1 ] ; sld:type [ sld:items cwl:File ; sld:type sld:array ] ; sld:type [ sld:items cwl:File ; sld:type sld:array ] . sld:type cwl:File . rdfs:comment "Best evalue plus this exponent that will be returned, default 0 (no range)" ; cwl:default 0 ; cwl:inputBinding [ CommandLineBinding:prefix "-r" ] ; sld:type xsd:int , sld:null . rdfs:label "Convert to fasta" ; cwl:in , , ; cwl:out ; cwl:run . sld:type xsd:string . cwl:in , , ; cwl:out ; cwl:run . rdfs:comment "Maximum size (in Gb) to count, files larger are split, default 10.0" ; cwl:default 1e+01 ; cwl:inputBinding [ CommandLineBinding:prefix "--max" ] ; sld:type xsd:float , sld:null . cwl:source ; cwl:valueFrom "${ return [self]; }" . cwl:source . cwl:source ; cwl:valueFrom "${ return [self]; }" . cwl:in , , , , ; cwl:out , , , , , , ; cwl:run . cwl:source . cwl:outputSource ; sld:type cwl:File . rdfs:comment "Number of fragment chunks to load in memory at once before processing, default 5000" ; cwl:default 5000 ; cwl:inputBinding [ CommandLineBinding:prefix "--frag_num" ] ; sld:type xsd:int , sld:null . cwl:default 5 ; sld:type xsd:int . cwl:outputSource ; sld:type cwl:File . rdfs:comment "Run for fast DNA/DNA remapping - not allowing introns" ; cwl:inputBinding [ CommandLineBinding:prefix "-fastMap" ] ; sld:type xsd:boolean , sld:null . rdfs:comment "max number of bps to process [default 100]" ; cwl:inputBinding [ CommandLineBinding:prefix "--bp_max" ] ; sld:type xsd:int , sld:null . cwl:source . cwl:source . cwl:outputSource ; sld:type cwl:File . cwl:source . cwl:source ; cwl:valueFrom "$(self).100.preprocess.removed.fna" . cwl:outputBinding [ CommandOutputBinding:glob "bleachsims.log" ] ; sld:type cwl:File . rdfs:comment "Output skipped fastq sequences" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName).skip" ] ; sld:type cwl:File . cwl:in , , , ; cwl:out ; cwl:run ; cwl:scatter ( ) ; cwl:scatterMethod ScatterMethod:dotproduct . cwl:default true . rdfs:comment "Size of hash, used to limit memory consumption for testing." ; cwl:default 1e+00 ; cwl:inputBinding [ CommandLineBinding:prefix "--size" ] ; sld:type xsd:float , sld:null . rdfs:comment "Output filtered similarity file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outFilterName)" ] ; sld:type cwl:File . a cwl:CommandLineTool ; rdfs:comment "sort text file base on given field(s)" ; rdfs:label "GNU sort" ; ns1:original_cwlVersion "v1.2" ; cwl:arguments ( [ CommandLineBinding:prefix "-T" ; cwl:valueFrom "$(runtime.tmpdir)" ] [ CommandLineBinding:prefix "-S" ; cwl:valueFrom "$(runtime.ram)M" ] ) ; cwl:arguments ( [ CommandLineBinding:prefix "-T" ; cwl:valueFrom "$(runtime.tmpdir)" ] [ CommandLineBinding:prefix "-S" ; cwl:valueFrom "$(runtime.ram)M" ] ) ; cwl:arguments ( [ CommandLineBinding:prefix "-T" ; cwl:valueFrom "$(runtime.tmpdir)" ] [ CommandLineBinding:prefix "-S" ; cwl:valueFrom "$(runtime.ram)M" ] ) ; cwl:baseCommand ( "sort" ) ; cwl:baseCommand ( "sort" ) ; cwl:baseCommand ( "sort" ) ; cwl:cwlVersion ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:inputs , , , , ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "sort.error" ; cwl:stdout "sort.log" . cwl:outputBinding [ CommandOutputBinding:glob "superblat.log" ] ; sld:type cwl:File . rdfs:comment "Optional input file, assembly coverage" ; cwl:inputBinding [ CommandLineBinding:prefix "--coverage" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--coverage" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--coverage" ] ; sld:type cwl:File , sld:null . cwl:in ; cwl:out , ; cwl:run . cwl:outputBinding [ CommandOutputBinding:glob "sims_annotate.log" ] ; sld:type cwl:File . a cwl:CommandLineTool ; rdfs:comment "cluster protein sequences\nuse max available cpus and memory\n>cdhit -n 5 -d 0 -T 0 -M 0 -c 0.9 -i -o \n" ; rdfs:label "CD-HIT" ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( [ CommandLineBinding:prefix "-M" ; cwl:valueFrom "0" ] [ CommandLineBinding:prefix "-T" ; cwl:valueFrom "0" ] [ CommandLineBinding:prefix "-d" ; cwl:valueFrom "0" ] ) ; cwl:baseCommand ( "cdhit" ) ; cwl:cwlVersion ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:inputs , , , ; cwl:outputs , , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "cdhit.error" ; cwl:stdout "cdhit.log" . a cwl:CommandLineTool ; rdfs:comment "Utility tool for various sequence file transformations.\n" ; rdfs:label "seqUtil" ; ns1:original_cwlVersion "v1.2" ; cwl:baseCommand ( "seqUtil" ) ; cwl:baseCommand ( "seqUtil" ) ; cwl:baseCommand ( "seqUtil" ) ; cwl:cwlVersion ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , , , , , , , , , ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "seqUtil.error" ; cwl:stdout "seqUtil.log" . rdfs:comment "Filename to place gc bins" ; cwl:inputBinding [ CommandLineBinding:prefix "--gc_percent_bin" ] ; sld:type xsd:string , sld:null . cwl:valueFrom "protein" . cwl:in , , , ; cwl:out , ; cwl:run . rdfs:comment "Fast stats, length and count only, for protein sequences" ; cwl:inputBinding [ CommandLineBinding:prefix "--fast" ] ; sld:type xsd:boolean , sld:null . cwl:outputBinding [ CommandOutputBinding:glob "drisee.stats" ] ; sld:type cwl:File . cwl:in , , ; cwl:out ; cwl:run . cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.output)" ] ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.output)" ] ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.output)" ] ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "uncluster_sims.log" ] ; sld:type cwl:File . cwl:in , ; cwl:out ; cwl:run . cwl:source . cwl:outputSource ; sld:type cwl:File . cwl:source ; cwl:valueFrom "$(self.basename).sort.tab" . cwl:outputBinding [ CommandOutputBinding:glob "drisee.log" ] ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "bleachsims.error" ] ; sld:type cwl:File . rdfs:comment "Output name" ; cwl:inputBinding [ CommandLineBinding:position 3 ] ; sld:type xsd:string . sld:type [ sld:items cwl:File ; sld:type sld:array ] . cwl:in , , ; cwl:out ; cwl:run . cwl:outputBinding [ CommandOutputBinding:glob "add_seq2sims.error" ] ; sld:type cwl:File . rdfs:label "Low quality filtering" ; cwl:in , , , , ; cwl:out , ; cwl:run . cwl:source . cwl:source . cwl:source ; cwl:valueFrom "$(self).080.adapter.trim.passed.fastq" . a cwl:Workflow ; rdfs:comment "protein - qc, preprocess, annotation, index, abundance" ; rdfs:label "metabarcode (gene amplicon) analysis for fastq files" ; ns1:original_cwlVersion "v1.0" ; Workflow:steps , , , , ; cwl:cwlVersion ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , , , , , , ; cwl:outputs , , , , , , , , , , , , , , ; cwl:requirements [ a cwl:ScatterFeatureRequirement ] ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:requirements [ a cwl:SubworkflowFeatureRequirement ] ; cwl:requirements [ a cwl:StepInputExpressionRequirement ] ; cwl:requirements [ a cwl:MultipleInputFeatureRequirement ] . cwl:source . a cwl:Workflow ; ns1:original_cwlVersion "v1.0" ; Workflow:steps , , , , , ; cwl:cwlVersion ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:inputs , , ; cwl:outputs , , , ; cwl:requirements [ a cwl:StepInputExpressionRequirement ] ; cwl:requirements [ a cwl:MultipleInputFeatureRequirement ] ; cwl:requirements [ a cwl:ScatterFeatureRequirement ] ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] . cwl:source . cwl:outputSource ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.output_prefix).seq.stats" ] ; sld:type cwl:File . sld:type xsd:int . cwl:outputBinding [ CommandOutputBinding:glob "index_sims_file_md5.error" ] ; sld:type cwl:File . rdfs:comment "Input file, sequence (fasta/fastq) or binary count hash (hash)." ; cwl:inputBinding [ CommandLineBinding:prefix "--input" ] ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "seq_length_stats.stats" ] ; sld:type cwl:File . a cwl:CommandLineTool ; rdfs:comment "fastq quality trimmer\n>fastq-mcf 'n/a' -S -k 0 -l --max-ns -q -w -o \n" ; rdfs:label "fastq-mcf" ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( [ CommandLineBinding:position 0 ; cwl:valueFrom "n/a" ] [ cwl:valueFrom "-S" ] [ CommandLineBinding:prefix "-k" ; cwl:valueFrom "0" ] ) ; cwl:baseCommand ( "fastq-mcf" ) ; cwl:cwlVersion ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , , , , , ; cwl:outputs , , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "fastq-mcf.error" ; cwl:stdout "fastq-mcf.log" . sld:type cwl:File . rdfs:comment "Profile type" ; cwl:inputBinding [ CommandLineBinding:prefix "-t" ] ; cwl:inputBinding [ CommandLineBinding:prefix "-t" ] ; cwl:inputBinding [ CommandLineBinding:prefix "-t" ] ; sld:type xsd:string . cwl:outputBinding [ CommandOutputBinding:glob "autoskewer.error" ] ; sld:type cwl:File . sld:type [ rdfs:label "none" ; sld:fields , ; sld:type sld:record ] . cwl:source ; cwl:valueFrom "$([self])" . rdfs:comment "Output stats in json format, default is tabbed text" ; cwl:inputBinding [ CommandLineBinding:prefix "--json" ] ; sld:type xsd:boolean , sld:null . rdfs:comment "Ignore commas in header ID" ; cwl:inputBinding [ CommandLineBinding:prefix "--ignore_comma" ] ; sld:type xsd:boolean , sld:null . cwl:source . cwl:source . cwl:source . cwl:source ; cwl:valueFrom "$(self).075.gc.bin" . rdfs:comment "Input file, sequence (fasta/fastq)" ; cwl:inputBinding [ CommandLineBinding:prefix "--input" ] ; sld:type cwl:File . rdfs:comment "output prefix, e.g. ${output_prefix}.seq.bins, ${output_prefix}.seq.stats" ; cwl:inputBinding [ CommandLineBinding:prefix "-out_prefix" ] ; sld:type xsd:string . rdfs:comment "drisee stat file" ; cwl:inputBinding [ CommandLineBinding:prefix "-drisee_stat" ] ; sld:type cwl:File , sld:null . a cwl:CommandLineTool ; rdfs:comment "Calculate statistics for fasta files.\nOutput fields:\\\n bp_count\n sequence_count\n average_length\n standard_deviation_length\n length_min\n length_max\n average_gc_content\n standard_deviation_gc_content\n average_gc_ratio\n standard_deviation_gc_ratio\n ambig_char_count\n ambig_sequence_count\n average_ambig_chars\n sequence_type\n" ; rdfs:label "sequence statistics" ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( [ CommandLineBinding:prefix "--type" ; cwl:valueFrom "${\n return inputs.sequences.basename.split(\".\").slice(-1)[0]\n}\n" ] ) ; cwl:baseCommand ( "seq_length_stats.py" ) ; cwl:cwlVersion ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , , , , , , , , ; cwl:outputs , , , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "seq_length_stats.error" ; cwl:stdout "seq_length_stats.stats" . cwl:outputBinding [ CommandOutputBinding:glob "kmer-tool.log" ] ; sld:type cwl:File . rdfs:comment "use SEP instead of non-blank to blank transition, default is tab" ; cwl:inputBinding [ CommandLineBinding:prefix "-t" ; cwl:valueFrom "$(\"\\u0009\")" ] ; cwl:inputBinding [ CommandLineBinding:prefix "-t" ; cwl:valueFrom "$(\"\\u0009\")" ] ; cwl:inputBinding [ CommandLineBinding:prefix "-t" ; cwl:valueFrom "$(\"\\u0009\")" ] ; sld:type xsd:string , sld:null . rdfs:comment "Input fasta format file" ; cwl:inputBinding [ CommandLineBinding:prefix "-i" ] ; sld:type cwl:File . cwl:source . rdfs:comment "Return fasta with each sequence truncated to 1024 bps" ; cwl:inputBinding [ CommandLineBinding:prefix "--subset_seqs" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--subset_seqs" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--subset_seqs" ] ; sld:type xsd:boolean , sld:null . cwl:source . rdfs:comment "Optional output trimmed log" ; cwl:inputBinding [ CommandLineBinding:prefix "-l" ] ; sld:type xsd:string , sld:null . cwl:outputSource ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "index_sims_file_md5.log" ] ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "drisee.error" ] ; sld:type cwl:File . cwl:outputSource ; sld:type cwl:File . cwl:source . cwl:in , , ; cwl:out ; cwl:run . cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.prefix).kmer.$(inputs.length).stats" ] ; sld:type cwl:File . rdfs:comment "Column position of query in sims file, default is 1" ; cwl:default 1 ; cwl:inputBinding [ CommandLineBinding:prefix "--position" ] ; sld:type xsd:int , sld:null . cwl:source . rdfs:comment "Compute nucleotide histogram for all reads" ; rdfs:label "Nucleotide Histogram" ; cwl:in , , ; cwl:out , ; cwl:run . cwl:outputSource ; sld:type cwl:File . rdfs:comment "md5 single copy gene file" ; cwl:inputBinding [ CommandLineBinding:prefix "--in_scg" ] ; sld:type cwl:File , sld:null . cwl:source . cwl:source ; cwl:valueFrom "${ return [self]; }" . cwl:outputBinding [ CommandOutputBinding:glob "uncluster_sims.error" ] ; sld:type cwl:File . rdfs:comment "kmer record array" ; sld:type [ sld:items [ sld:fields , ; sld:type sld:record ] ; sld:type sld:array ] . cwl:source . cwl:outputSource ; sld:type cwl:File . cwl:source ; cwl:valueFrom "$(self).700.annotation.source.stats" . cwl:source . cwl:source ; cwl:valueFrom "${ return [self]; }" . cwl:source . cwl:source ; cwl:valueFrom "$(self).650.aa.expand.protein" . rdfs:comment "stats tabbed file" ; cwl:inputBinding [ CommandLineBinding:prefix "-seq_stat" ] ; sld:type cwl:File . cwl:in , , , , ; cwl:out ; cwl:run . rdfs:comment "Output expanded LCA file (protein and rna mode)" ; cwl:inputBinding [ CommandLineBinding:prefix "--out_lca" ] ; sld:type xsd:string . cwl:source . cwl:source ; cwl:valueFrom "$(self[0]).700.annotation.sims.filter.seq" . rdfs:comment "Output name" ; cwl:inputBinding [ CommandLineBinding:prefix "-o" ] ; sld:type xsd:string . rdfs:comment "Database fasta format file" ; cwl:inputBinding [ CommandLineBinding:position 1 ] ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.output)" ] ; sld:type cwl:File . rdfs:comment "Transform fasta to ID list, sorted by ID" ; cwl:inputBinding [ CommandLineBinding:prefix "--sortbyid2id" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--sortbyid2id" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--sortbyid2id" ] ; sld:type xsd:boolean , sld:null . sld:type [ sld:items cwl:File ; sld:type sld:array ] . cwl:source . cwl:in , , , ; cwl:out , ; cwl:run . cwl:outputBinding [ CommandOutputBinding:glob "kmer-tool.error" ] ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "cdhit.log" ] ; sld:type cwl:File . rdfs:comment "Maxmium N-calls in a read, default 5" ; cwl:default 5 ; cwl:inputBinding [ CommandLineBinding:prefix "--max-ns" ] ; sld:type xsd:int , sld:null . cwl:outputSource ; sld:type cwl:File . cwl:source . cwl:source ; cwl:valueFrom "${ return [self]; }" . cwl:outputSource ; sld:type cwl:File . cwl:inputBinding [ CommandLineBinding:prefix "--stats" ] ; sld:type cwl:File , sld:null . cwl:source ; cwl:valueFrom "$(self[0].basename).sort" . cwl:source ; cwl:valueFrom "$(self).550.cluster.aa90.mapping" . sld:type cwl:File . cwl:source . rdfs:comment "Output filtered similarity blast-m8 file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; sld:type cwl:File . cwl:in , , ; cwl:out , , , ; cwl:run . rdfs:comment "Concatenated file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; sld:type cwl:File . cwl:in , , , ; cwl:out , , ; cwl:run . rdfs:comment "Type of sequences data in input file, rna or protein" ; cwl:default "protein" ; cwl:inputBinding [ CommandLineBinding:prefix "--format" ] ; sld:type xsd:string . rdfs:comment "Output index file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "run_FragGeneScan.log" ] ; sld:type cwl:File . cwl:source . cwl:source . cwl:in , , ; cwl:out , ; cwl:run . cwl:outputSource ; sld:type cwl:File . sld:type cwl:File . cwl:source ; cwl:valueFrom "$(self).650.aa.sims.filter" . cwl:in , , ; cwl:out ; cwl:run . cwl:valueFrom "source" . cwl:source . cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outPrefix).qc.summary" ] ; sld:type cwl:File . cwl:source . cwl:outputBinding [ CommandOutputBinding:glob "cdhit.error" ] ; sld:type cwl:File . cwl:outputSource ; sld:type cwl:File . rdfs:label "Convert to fasta" ; cwl:in , , ; cwl:out ; cwl:run . cwl:default true . cwl:outputSource ; sld:type cwl:File . cwl:in , , ; cwl:out ; cwl:run . cwl:outputBinding [ CommandOutputBinding:glob "format_cluster.log" ] ; sld:type cwl:File . rdfs:comment "Output trimmed sequences" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; sld:type cwl:File . rdfs:comment "Output name" ; cwl:inputBinding [ CommandLineBinding:prefix "-o" ] ; sld:type xsd:string . rdfs:comment "Input fasta format file" ; cwl:inputBinding [ CommandLineBinding:prefix "--genome" ] ; sld:type cwl:File . a cwl:CommandLineTool ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( [ CommandLineBinding:prefix "-kmer_lens" ; cwl:valueFrom "${\n return inputs.kmer.map( \n function(r){ return r.length }\n ).join();\n}\n" ] [ CommandLineBinding:prefix "-kmer_stats" ; cwl:valueFrom "${\n return inputs.kmer.map( \n function(r){ return r.file.path }\n ).join();\n}\n" ] ) ; cwl:baseCommand ( "format_qc_stats.pl" ) ; cwl:cwlVersion ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:inputs , , , , , ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "format_qc_stats.error" ; cwl:stdout "format_qc_stats.stats" . cwl:in , , , , , ; cwl:out , , ; cwl:run . cwl:source . rdfs:comment "Input expanded sims file" ; sld:type [ cwl:inputBinding [ CommandLineBinding:prefix "-i" ] ; sld:items cwl:File ; sld:type sld:array ] ; sld:type [ cwl:inputBinding [ CommandLineBinding:prefix "-i" ] ; sld:items cwl:File ; sld:type sld:array ] ; sld:type [ cwl:inputBinding [ CommandLineBinding:prefix "-i" ] ; sld:items cwl:File ; sld:type sld:array ] . cwl:source . cwl:source . rdfs:comment "Output file." ; cwl:inputBinding [ CommandLineBinding:prefix "--output" ] ; sld:type xsd:string . cwl:default 15 ; sld:type xsd:int . rdfs:comment "Output trimmed sequences" ; cwl:inputBinding [ CommandLineBinding:prefix "-o" ] ; sld:type xsd:string . cwl:source . cwl:default 15 ; sld:type xsd:int . rdfs:comment "Return fasta with each sequence truncated to 1024 bps" ; cwl:inputBinding [ CommandLineBinding:prefix "--bowtie_truncate" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--bowtie_truncate" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--bowtie_truncate" ] ; sld:type xsd:boolean , sld:null . cwl:outputBinding [ CommandOutputBinding:glob "autoskewer.log" ] ; sld:type cwl:File . rdfs:comment "Output .mapping format file" ; cwl:inputBinding [ CommandLineBinding:prefix "--output" ] ; sld:type xsd:string . cwl:source . rdfs:comment "Word length, default 5" ; cwl:default 5 ; cwl:inputBinding [ CommandLineBinding:prefix "-n" ] ; sld:type xsd:int , sld:null . cwl:source . rdfs:comment "Input cluster mapping files" ; sld:type [ cwl:inputBinding [ CommandLineBinding:prefix "-c" ] ; sld:items cwl:File ; sld:type sld:array ] . rdfs:comment "Output merged sims and seq file" ; cwl:outputBinding [ CommandOutputBinding:glob "$([inputs.outName])" ] ; sld:type [ sld:items cwl:File ; sld:type sld:array ] . cwl:outputBinding [ CommandOutputBinding:glob "fastq-mcf.error" ] ; sld:type cwl:File . sld:type [ sld:items cwl:File ; sld:type sld:array ] . rdfs:comment "Output abundance profile file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; sld:type cwl:File . cwl:source . cwl:outputSource ; sld:type cwl:File . sld:type cwl:File . cwl:source ; cwl:valueFrom "$(self).550.cluster.aa90.faa" . rdfs:comment "Window-size for quality trimming, default 10" ; cwl:default 10 ; cwl:inputBinding [ CommandLineBinding:prefix "-w" ] ; sld:type xsd:int , sld:null . a cwl:CommandLineTool ; rdfs:comment "filter similarity file by E-value and number of hits\n>bleachsims -s -o -m 20 -r 0 -c 3\n" ; rdfs:label "bleachsims" ; ns1:original_cwlVersion "v1.0" ; cwl:baseCommand ( "bleachsims" ) ; cwl:cwlVersion ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , , , , , , ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "bleachsims.error" ; cwl:stdout "bleachsims.log" . rdfs:comment "Verbose logging mode" ; cwl:inputBinding [ CommandLineBinding:prefix "--verbose" ] ; sld:type xsd:boolean , sld:null . sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "sims_annotate.error" ] ; sld:type cwl:File . cwl:source . a cwl:Workflow ; rdfs:comment "abundace profiles from annotated files, for protein and/or rna" ; rdfs:label "abundance" ; ns1:original_cwlVersion "v1.0" ; Workflow:steps , , ; cwl:cwlVersion ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , , , , , , ; cwl:outputs , , ; cwl:requirements [ a cwl:ScatterFeatureRequirement ] ; cwl:requirements [ a cwl:StepInputExpressionRequirement ] ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:requirements [ a cwl:MultipleInputFeatureRequirement ] . cwl:outputSource ; sld:type cwl:File . a cwl:CommandLineTool ; rdfs:comment "Script to use jellyfish to get kmer information\nInput:\\ fasta/fastq file\nOutput:\\ kmer information, one of:\\\n 1. hash:\\ binary hash of counts\n 2. stats:\\ summary stats\n 3. dump:\\ profile (kmer seq - count)\n 4. histo:\\ histogram (count - abundance)\n 5. histo ranked:\\ count, abundance, count*abundance, reverse-sum(abundance), reverse-sum(count*abundance), ratio-to-largest\n" ; rdfs:label "calcualate kmer bins" ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( "--ranked" [ CommandLineBinding:prefix "--procs" ; cwl:valueFrom "$(runtime.cores)" ] [ CommandLineBinding:prefix "--type" ; cwl:valueFrom "${\n return inputs.sequences.basename.split(\".\").slice(-1)[0]\n}\n" ] [ CommandLineBinding:prefix "--tmpdir" ; cwl:valueFrom "$(runtime.outdir)" ] [ CommandLineBinding:prefix "--output" ; cwl:valueFrom "$(inputs.prefix).kmer.$(inputs.length).stats" ] ) ; cwl:baseCommand ( "kmer-tool" ) ; cwl:cwlVersion ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , , , , , ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "kmer-tool.error" ; cwl:stdout "kmer-tool.log" . a cwl:CommandLineTool ; rdfs:comment "hidden Markov model for predicting prokaryotic coding regions\n>run_FragGeneScan.pl --genome --out --complete 0 --train 454_30\n" ; rdfs:label "FragGeneScan" ; ns1:original_cwlVersion "v1.0" ; cwl:baseCommand ( "run_FragGeneScan.pl" ) ; cwl:cwlVersion ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , , , ; cwl:outputs , , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "run_FragGeneScan.error" ; cwl:stdout "run_FragGeneScan.log" . cwl:in , ; cwl:out ; cwl:run . a cwl:CommandLineTool ; rdfs:comment "Concatenate FILE(s) to standard output" ; rdfs:label "GNU cat" ; ns1:original_cwlVersion "v1.2" ; cwl:baseCommand ( "cat" ) ; cwl:baseCommand ( "cat" ) ; cwl:cwlVersion ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , ; cwl:outputs , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "cat.error" ; cwl:stdout "$(inputs.outName)" . cwl:outputBinding [ CommandOutputBinding:glob "consensus.error" ] ; sld:type cwl:File . rdfs:comment "Transform fasta to tabbed" ; cwl:inputBinding [ CommandLineBinding:prefix "--fasta2tab" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--fasta2tab" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--fasta2tab" ] ; sld:type xsd:boolean , sld:null . rdfs:comment "Output stats file name, if not called prints to STDOUT" ; cwl:inputBinding [ CommandLineBinding:prefix "--output" ] ; sld:type xsd:string , sld:null . cwl:source . sld:type [ sld:items cwl:File ; sld:type sld:array ] . cwl:source ; cwl:valueFrom "$(self).650.aa.expand.lca" . rdfs:comment "Output format, one of [hash, stats, dump, histo], default histo" ; cwl:default "histo" ; cwl:inputBinding [ CommandLineBinding:prefix "--format" ] ; sld:type xsd:string , sld:null . rdfs:comment "only apply cutoff filter" ; cwl:inputBinding [ CommandLineBinding:prefix "--min_hit_only" ] ; sld:type xsd:boolean , sld:null . sld:type [ sld:items cwl:File ; sld:type sld:array ] . rdfs:comment "len bin file" ; cwl:inputBinding [ CommandLineBinding:prefix "-seq_lens" ] ; sld:type cwl:File . cwl:source . rdfs:comment "Input .clstr format file" ; cwl:inputBinding [ CommandLineBinding:prefix "--input" ] ; sld:type cwl:File . cwl:default 5 ; sld:type xsd:int . rdfs:comment "Input similarity hit files" ; sld:type [ cwl:inputBinding [ CommandLineBinding:prefix "-i" ] ; sld:items cwl:File ; sld:type sld:array ] . a cwl:CommandLineTool ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( [ CommandLineBinding:position 2 ; cwl:valueFrom "drisee.stats" ] "--verbose" "--filter_seq" [ CommandLineBinding:prefix "--processes" ; cwl:valueFrom "$(runtime.cores)" ] [ CommandLineBinding:prefix "--tmp_dir" ; cwl:valueFrom "$(runtime.tmpdir)" ] [ CommandLineBinding:prefix "--seq_type" ; cwl:valueFrom "${\n return inputs.sequences.basename.split(\".\").slice(-1)[0]\n } \n" ] ) ; cwl:baseCommand ( "drisee" ) ; cwl:cwlVersion ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:inputs ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "drisee.error" ; cwl:stdout "drisee.log" . cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outPrefix).qc.stats" ] ; sld:type cwl:File . a cwl:CommandLineTool ; rdfs:comment "create expanded annotated sims files from input md5 sim file and m5nr db\nsims_annotate.pl --verbose --in_sim --in_scg --ann_file --format --out_filter --out_expand -out_lca --frag_num 5000\n" ; rdfs:label "annotate sims" ; ns1:original_cwlVersion "v1.0" ; cwl:baseCommand ( "sims_annotate.pl" ) ; cwl:cwlVersion ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:inputs , , , , , , , , ; cwl:outputs , , , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "sims_annotate.error" ; cwl:stdout "sims_annotate.log" . rdfs:comment "Output filtered sim file" ; cwl:inputBinding [ CommandLineBinding:prefix "--out_filter" ] ; sld:type xsd:string . cwl:in , , , ; cwl:out ; cwl:run ; cwl:scatter ( ) ; cwl:scatterMethod ScatterMethod:dotproduct . cwl:source . sld:type cwl:File , sld:null . cwl:outputBinding [ CommandOutputBinding:glob "add_seq2sims.log" ] ; sld:type cwl:File . a cwl:CommandLineTool ; ns1:original_cwlVersion "v1.0" ; cwl:baseCommand ( "format_seq_stats.pl" ) ; cwl:cwlVersion ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:inputs , , , ; cwl:outputs , , ; cwl:stderr "format_seq_stats.error" ; cwl:stdout "format_seq_stats.stats" . cwl:outputBinding [ CommandOutputBinding:glob "fastq-mcf.log" ] ; sld:type cwl:File . cwl:in , , , , ; cwl:out ; cwl:run . cwl:source ; cwl:valueFrom "$(self).100.preprocess.passed.fna" . cwl:outputSource ; sld:type cwl:File . cwl:source . cwl:outputBinding [ CommandOutputBinding:glob "cat.error" ] ; cwl:outputBinding [ CommandOutputBinding:glob "cat.error" ] ; sld:type cwl:File . rdfs:comment "Output sequence file" ; cwl:inputBinding [ CommandLineBinding:prefix "--output" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--output" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--output" ] ; sld:type xsd:string . cwl:source . cwl:valueFrom "$([\"1,1\"])" . rdfs:comment "Input similarity blast-m8 file" ; cwl:inputBinding [ CommandLineBinding:prefix "-s" ] ; sld:type cwl:File . rdfs:comment "Minimum remaining sequence length, default 50" ; cwl:default 50 ; cwl:inputBinding [ CommandLineBinding:prefix "-l" ] ; sld:type xsd:int , sld:null . cwl:outputBinding [ CommandOutputBinding:glob "run_FragGeneScan.error" ] ; sld:type cwl:File . cwl:source ; cwl:valueFrom "$(self).superblat.sims.raw" . rdfs:comment "Files to sort" ; cwl:inputBinding [ CommandLineBinding:position 2 ] ; cwl:inputBinding [ CommandLineBinding:position 2 ] ; cwl:inputBinding [ CommandLineBinding:position 2 ] ; sld:type [ sld:items cwl:File ; sld:type sld:array ] ; sld:type [ sld:items cwl:File ; sld:type sld:array ] ; sld:type [ sld:items cwl:File ; sld:type sld:array ] . rdfs:comment "Training model to use, default is 454_30" ; cwl:default "454_30" ; cwl:inputBinding [ CommandLineBinding:prefix "--train" ] ; sld:type xsd:string , sld:null . rdfs:comment "Output abundance profile" ; cwl:inputBinding [ CommandLineBinding:prefix "-o" ] ; cwl:inputBinding [ CommandLineBinding:prefix "-o" ] ; cwl:inputBinding [ CommandLineBinding:prefix "-o" ] ; sld:type xsd:string . cwl:source . rdfs:comment "Transform fastq to fasta" ; cwl:inputBinding [ CommandLineBinding:prefix "--fastq2fasta" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--fastq2fasta" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--fastq2fasta" ] ; sld:type xsd:boolean , sld:null . cwl:source ; cwl:valueFrom "$(self).700.annotation.lca.abundance" . cwl:source . cwl:source ; cwl:valueFrom "$(self).uncluster.sims" . rdfs:comment "Output tab separated similarity file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; sld:type cwl:File . sld:type xsd:string . rdfs:comment "Input tabbed sequence file" ; cwl:inputBinding [ CommandLineBinding:prefix "--seq_file" ] ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.output_prefix).seq.bins" ] ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "consensus.log" ] ; sld:type cwl:File . cwl:valueFrom "${ return [6, 15]; }" . cwl:source . rdfs:comment "Input sequence file" ; cwl:inputBinding [ CommandLineBinding:prefix "-i" ] ; sld:type cwl:File . cwl:default true . cwl:valueFrom "lca" . cwl:source ; cwl:valueFrom "${ return [self]; }" . cwl:valueFrom "$([\"2,2\"])" . cwl:outputBinding [ CommandOutputBinding:glob "seq_length_stats.error" ] ; sld:type cwl:File . rdfs:label "Adapter Trimming" ; cwl:in , ; cwl:out ; cwl:run . rdfs:comment "Input similarity file" ; cwl:inputBinding [ CommandLineBinding:prefix "--in_sim" ] ; sld:type cwl:File . cwl:source . rdfs:comment "Remove all evalues with an exponent lower than cutoff, default 3" ; cwl:default 3 ; cwl:inputBinding [ CommandLineBinding:prefix "-c" ] ; sld:type xsd:int , sld:null . cwl:in , ; cwl:out ; cwl:run . cwl:outputBinding [ CommandOutputBinding:glob "format_qc_stats.error" ] ; sld:type cwl:File . rdfs:comment "Output cluster mapping file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName).clstr" ] ; sld:type cwl:File . cwl:source ; cwl:valueFrom "$(self).350.genecalling.coding" . cwl:outputBinding [ CommandOutputBinding:glob "format_cluster.error" ] ; sld:type cwl:File . cwl:source ; cwl:valueFrom "$(self).100.preprocess.fastq" . sld:type cwl:File . cwl:source . cwl:source ; cwl:valueFrom "$(self).075.consensus.stats" . cwl:outputBinding [ CommandOutputBinding:glob "sort.log" ] ; cwl:outputBinding [ CommandOutputBinding:glob "sort.log" ] ; cwl:outputBinding [ CommandOutputBinding:glob "sort.log" ] ; sld:type cwl:File . cwl:source ; cwl:valueFrom "$(self).075.qc" . rdfs:comment "List of sequences to subset input by, required with subsetSeqs option" ; cwl:inputBinding [ CommandLineBinding:prefix "--list" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--list" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--list" ] ; sld:type cwl:File , sld:null . cwl:in , ; cwl:out ; cwl:run . cwl:source . rdfs:comment "Output unclustered similarity" ; cwl:inputBinding [ CommandLineBinding:prefix "-o" ] ; sld:type xsd:string . cwl:valueFrom "histo" . a cwl:Workflow ; rdfs:comment "Proteins - predict, cluster, identify, annotate" ; rdfs:label "protein annotation" ; ns1:original_cwlVersion "v1.0" ; Workflow:steps , , , , , , , ; cwl:cwlVersion ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , , , , , ; cwl:outputs , , , , , , ; cwl:requirements [ a cwl:ScatterFeatureRequirement ] ; cwl:requirements [ a cwl:StepInputExpressionRequirement ] ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:requirements [ a cwl:MultipleInputFeatureRequirement ] . cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; sld:type cwl:File , sld:null . rdfs:comment "Output expanded LCA file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outLcaName)" ] ; sld:type cwl:File . rdfs:comment "drisee info file" ; cwl:inputBinding [ CommandLineBinding:prefix "-drisee_info" ] ; sld:type cwl:File , sld:null . a cwl:CommandLineTool ; rdfs:comment "re-formats cd-hit .clstr file into mg-rast .mapping file\n>format_cluster.pl --input --output \n" ; rdfs:label "cluster file reformat" ; ns1:original_cwlVersion "v1.0" ; cwl:baseCommand ( "format_cluster.pl" ) ; cwl:cwlVersion ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "format_cluster.error" ; cwl:stdout "format_cluster.log" . cwl:outputSource ; sld:type cwl:File . cwl:in , , , , , , ; cwl:out , , ; cwl:run . cwl:default 6 ; sld:type sld:null ; sld:type [ sld:items xsd:int ; sld:type sld:array ] . rdfs:comment "Quality threshold causing base removal, default 15" ; cwl:default 15 ; cwl:inputBinding [ CommandLineBinding:prefix "-q" ] ; sld:type xsd:int , sld:null . rdfs:comment "only apply cutoff filter" ; cwl:inputBinding [ CommandLineBinding:prefix "--eval_only" ] ; sld:type xsd:boolean , sld:null . cwl:outputSource ; sld:type cwl:File . rdfs:comment "max number of seqs to process (for kmer entropy)" ; cwl:default 100000 ; cwl:inputBinding [ CommandLineBinding:prefix "--seq_max" ] ; sld:type xsd:int , sld:null . cwl:outputBinding [ CommandOutputBinding:glob "sims_abundance.log" ] ; cwl:outputBinding [ CommandOutputBinding:glob "sims_abundance.log" ] ; cwl:outputBinding [ CommandOutputBinding:glob "sims_abundance.log" ] ; sld:type cwl:File . cwl:default 9e-01 ; sld:type xsd:float , sld:null . cwl:source ; cwl:valueFrom "$(self).650.superblat.sims" . cwl:outputBinding [ CommandOutputBinding:glob "sort.error" ] ; cwl:outputBinding [ CommandOutputBinding:glob "sort.error" ] ; cwl:outputBinding [ CommandOutputBinding:glob "sort.error" ] ; sld:type cwl:File . cwl:source . cwl:outputSource ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "format_seq_stats.error" ] ; sld:type cwl:File . cwl:source . rdfs:comment "consensus stat file" ; cwl:inputBinding [ CommandLineBinding:prefix "-consensus" ] ; sld:type cwl:File , sld:null . rdfs:comment "start a key at POS1, end it at POS2 (origin 1)" ; cwl:inputBinding [ CommandLineBinding:position 1 ] ; cwl:inputBinding [ CommandLineBinding:position 1 ] ; cwl:inputBinding [ CommandLineBinding:position 1 ] ; sld:type [ cwl:inputBinding [ CommandLineBinding:prefix "-k" ] ; sld:items xsd:string ; sld:type sld:array ] ; sld:type [ cwl:inputBinding [ CommandLineBinding:prefix "-k" ] ; sld:items xsd:string ; sld:type sld:array ] ; sld:type [ cwl:inputBinding [ CommandLineBinding:prefix "-k" ] ; sld:items xsd:string ; sld:type sld:array ] . cwl:in , , , , ; cwl:out , ; cwl:run . cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.gcBin)" ] ; sld:type cwl:File , sld:null . rdfs:comment "Output index" ; cwl:inputBinding [ CommandLineBinding:prefix "--out_file" ] ; sld:type xsd:string . a cwl:CommandLineTool ; rdfs:comment "index m8 format blast file by 2nd column (sorted required)\nreturn: value, seek, length for each record\n>index_sims_file_md5 --in_file --out_file --md5_num \n" ; rdfs:label "index sims by md5" ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( [ cwl:valueFrom "--verbose" ] ) ; cwl:baseCommand ( "index_sims_file_md5" ) ; cwl:cwlVersion ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:inputs , , ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "index_sims_file_md5.error" ; cwl:stdout "index_sims_file_md5.log" . rdfs:comment "Output merged sims and seq" ; cwl:inputBinding [ CommandLineBinding:prefix "--out_sim" ] ; sld:type xsd:string . sld:type xsd:string . rdfs:comment "Optional output trimmed log file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outLog)" ] ; sld:type cwl:File , sld:null . cwl:outputSource ; sld:type cwl:File . sld:type cwl:File . sld:type [ sld:items cwl:File ; sld:type sld:array ] . cwl:default true . rdfs:comment "Sort fasta file by ID" ; cwl:inputBinding [ CommandLineBinding:prefix "--sortbyid" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--sortbyid" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--sortbyid" ] ; sld:type xsd:boolean , sld:null . rdfs:comment "Query fasta format file" ; cwl:inputBinding [ CommandLineBinding:position 2 ] ; sld:type cwl:File . cwl:outputSource ; sld:type cwl:File . cwl:source ; cwl:valueFrom "$(self.basename).index" . cwl:in , ; cwl:out ; cwl:run . cwl:inputBinding [ CommandLineBinding:position 1 ] ; sld:type cwl:File . rdfs:comment "Output name" ; cwl:inputBinding [ CommandLineBinding:prefix "--out" ] ; sld:type xsd:string . cwl:source . rdfs:comment "gc bin file" ; cwl:inputBinding [ CommandLineBinding:prefix "-seq_gc" ] ; sld:type cwl:File . rdfs:comment "Transform fasta to tabbed, sorted by ID" ; cwl:inputBinding [ CommandLineBinding:prefix "--sortbyid2tab" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--sortbyid2tab" ] ; cwl:inputBinding [ CommandLineBinding:prefix "--sortbyid2tab" ] ; sld:type xsd:boolean , sld:null . rdfs:comment "Guess sequence type [wgs|amplicon] from kmer entropy" ; cwl:inputBinding [ CommandLineBinding:prefix "--seq_type" ] ; sld:type xsd:boolean , sld:null . rdfs:comment "Filename to place length bins [default is no output]" ; cwl:inputBinding [ CommandLineBinding:prefix "--length_bin" ] ; sld:type xsd:string , sld:null . cwl:outputSource ; sld:type cwl:File . rdfs:comment "merge only, the input files are assumed to be pre-sorted" ; cwl:inputBinding [ CommandLineBinding:prefix "-m" ] ; cwl:inputBinding [ CommandLineBinding:prefix "-m" ] ; cwl:inputBinding [ CommandLineBinding:prefix "-m" ] ; sld:type xsd:boolean , sld:null . rdfs:comment "Optional input file(s), cluster mapping" ; sld:type sld:null ; sld:type [ cwl:inputBinding [ CommandLineBinding:prefix "--cluster" ] ; sld:items cwl:File ; sld:type sld:array ] ; sld:type [ cwl:inputBinding [ CommandLineBinding:prefix "--cluster" ] ; sld:items cwl:File ; sld:type sld:array ] ; sld:type [ cwl:inputBinding [ CommandLineBinding:prefix "--cluster" ] ; sld:items cwl:File ; sld:type sld:array ] . rdfs:comment "Prefix for output file(s)" ; cwl:default "qc" ; sld:type xsd:string , sld:null . a cwl:CommandLineTool ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( [ cwl:valueFrom "--verbose" ] [ CommandLineBinding:prefix "--type" ; cwl:valueFrom "${\n return inputs.sequences.basename.split(\".\").slice(-1)[0]\n} \n" ] ) ; cwl:baseCommand ( "consensus.py" ) ; cwl:cwlVersion ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , , , ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "consensus.error" ; cwl:stdout "consensus.log" . cwl:source . rdfs:comment "Output unclustered similarity file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; sld:type [ sld:items cwl:File ; sld:type sld:array ] . sld:type [ sld:items cwl:File ; sld:type sld:array ] . rdfs:comment "BerkelyDB of condensed M5NR" ; cwl:inputBinding [ CommandLineBinding:prefix "--ann_file" ] ; sld:type cwl:File . sld:type cwl:File . rdfs:comment "Number of chunks to load in memory at once before processing, default is 5000" ; cwl:default 5000 ; cwl:inputBinding [ CommandLineBinding:prefix "--md5_num" ] ; sld:type xsd:int , sld:null . sld:type xsd:string . cwl:source . cwl:source ; cwl:valueFrom "$(self.basename).superblat.sims" . sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "superblat.error" ] ; sld:type cwl:File . cwl:source ; cwl:valueFrom "$(self).075.seq.stats" . rdfs:comment "coverage stat file" ; cwl:inputBinding [ CommandLineBinding:prefix "-coverage" ] ; sld:type cwl:File , sld:null . rdfs:comment "Output trimmed fastq sequences" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; sld:type cwl:File . cwl:outputSource ; sld:type cwl:File . rdfs:comment "Percent identity threshold, default 0.9" ; cwl:default 9e-01 ; cwl:inputBinding [ CommandLineBinding:prefix "-c" ] ; sld:type xsd:float , sld:null . rdfs:comment "Output expanded protein sim file (protein mode only)" ; cwl:inputBinding [ CommandLineBinding:prefix "--out_expand" ] ; sld:type xsd:string . rdfs:comment "Input fastq sequence file" ; cwl:inputBinding [ CommandLineBinding:position 1 ] ; sld:type cwl:File . cwl:source . sld:type [ sld:items cwl:File ; sld:type sld:array ] . rdfs:comment "Input file, sequence (fasta/fastq)." ; cwl:inputBinding [ CommandLineBinding:prefix "--input" ] ; sld:type cwl:File . rdfs:comment "write result to FILE instead of standard output" ; cwl:inputBinding [ CommandLineBinding:prefix "-o" ] ; cwl:inputBinding [ CommandLineBinding:prefix "-o" ] ; cwl:inputBinding [ CommandLineBinding:prefix "-o" ] ; sld:type xsd:string . rdfs:comment "Output name" ; cwl:inputBinding [ CommandLineBinding:prefix "-o" ] ; sld:type xsd:string . cwl:outputSource ; sld:type cwl:File . rdfs:comment "Input similarity blast-m8 file" ; cwl:inputBinding [ CommandLineBinding:prefix "--in_sim" ] ; sld:type cwl:File . cwl:source . sld:type xsd:string . cwl:source ; cwl:valueFrom "$(self).075.qc" . cwl:valueFrom "md5" . cwl:source . sld:type xsd:string . a cwl:CommandLineTool ; rdfs:comment "detect and trim adapter sequences from reads\n>autoskewer.py -t -i -o -l \n" ; rdfs:label "autoskewer" ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( [ CommandLineBinding:prefix "-t" ; cwl:valueFrom "$(runtime.tmpdir)" ] ) ; cwl:baseCommand ( "autoskewer.py" ) ; cwl:cwlVersion ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.04" ] ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:inputs , , ; cwl:outputs , , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:requirements [ a cwl:InitialWorkDirRequirement ; cwl:listing "$(inputs.input)" ] ; cwl:stderr "autoskewer.error" ; cwl:stdout "autoskewer.log" . cwl:source ; cwl:valueFrom "$(self).075.length.bin" . rdfs:comment "Output .ffn (dna) file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName).ffn" ] ; sld:type cwl:File . cwl:outputSource ; sld:type cwl:File . rdfs:comment "Minimum number of hits per query, default 20" ; cwl:default 20 ; cwl:inputBinding [ CommandLineBinding:prefix "-m" ] ; sld:type xsd:int , sld:null . cwl:source ; cwl:valueFrom "$(self.basename).seq" . a cwl:CommandLineTool ; rdfs:comment "expand out similarity file (blast m8) by turning each cluster seed hit into a hit per cluster member\n>uncluster_sims.py --cfile --position \n" ; rdfs:label "uncluster sims" ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( [ cwl:valueFrom "--verbose" ] [ CommandLineBinding:prefix "--db" ; cwl:valueFrom "$(runtime.tmpdir)" ] ) ; cwl:baseCommand ( "uncluster_sims.py" ) ; cwl:cwlVersion ; cwl:hints [ a ; LoadListingRequirement:loadListing "deep_listing" ] ; cwl:hints [ a ; NetworkAccess:networkAccess true ] ; cwl:hints [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgrast/pipeline:4.03" ] ; cwl:inputs , , , ; cwl:outputs , , ; cwl:requirements [ a cwl:InlineJavascriptRequirement ] ; cwl:stderr "uncluster_sims.error" ; cwl:stdout "uncluster_sims.log" . cwl:outputSource ; sld:type cwl:File . rdfs:comment "Output .mapping format file" ; cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.outName)" ] ; sld:type cwl:File .