#@UGENE_WORKFLOW
#The workflow takes FASTQ files as input and process this data as follows:
#    - Filters the input sequencing reads by the CASAVA header (for Illumina platform).
#    - Cuts of adapter sequences.
#    - Trims the sequencing reads by quality.
#    - Merges different FASTQ files.
#    - Maps the sequencing reads by TopHat.
#
#The workflow also performs quality control of the data with FastQC: first, on the input FASTQ files and, second, after the pre-processing step.
#
#Besides intermediate files and FastQC reports, the final result is a BAM file with aligned reads and other TopHat output files.
#



workflow "Processing of raw RNA-Seq single-end reads with mapping"{

    CASAVAFilter {
        type:CASAVAFilter;
        name:"Filter Reads by CASAVA Header";
        custom-dir:filtered_fastq;
        out-mode:1;
    }
    QualityTrim {
        type:QualityTrim;
        name:"Trim Reads by Quality";
        custom-dir:filtered_fastq;
        len-id:10;
        qual-id:20;
    }
    get-file-list {
        type:get-file-list;
        name:"Read FASTQ Files with Reads";
        url-in {
            dataset:"Dataset 1";
        }
    }
    MergeFastq {
        type:MergeFastq;
        name:"Merge FASTQ Files";
        custom-dir:filtered_fastq;
        out-name:in_1.fq;
    }
    tophat {
        type:tophat;
        name:"Align Reads with TopHat";
        bowtie-version:1;
        mate-inner-distance:170;
        no-novel-junctions:true;
        out-dir:output;
    }
    CutAdaptFastq {
        type:CutAdaptFastq;
        name:"Cut Adapter";
    }
    fastqc {
        type:fastqc;
        name:"FastQC Quality Control";
        out-mode:1;
    }
    fastqc-1 {
        type:fastqc;
        name:"FastQC Quality Control";
        out-mode:1;
    }

    .actor-bindings {
        MergeFastq.out-file->tophat.in-sequence
        MergeFastq.out-file->fastqc-1.in-file
        QualityTrim.out-file->MergeFastq.in-file
        CutAdaptFastq.out-file->QualityTrim.in-file
        CASAVAFilter.out-file->CutAdaptFastq.in-file
        get-file-list.out-url->CASAVAFilter.in-file
        get-file-list.out-url->fastqc.in-file
    }

    get-file-list.url->CASAVAFilter.in-file.url
    CutAdaptFastq.url->QualityTrim.in-file.url
    QualityTrim.url->MergeFastq.in-file.url
    MergeFastq.url->tophat.in-sequence.in-url
    CASAVAFilter.url->CutAdaptFastq.in-file.url
    get-file-list.url->fastqc.in-file.url
    MergeFastq.url->fastqc-1.in-file.url

    .meta {
        parameter-aliases {
            CASAVAFilter.out-mode {
                alias:out_mode;
            }
            get-file-list.url-in {
                alias:in;
            }
            tophat.bowtie-index-basename {
                alias:idx_name;
            }
            tophat.bowtie-index-dir {
                alias:idx_dir;
            }
            tophat.out-dir {
                alias:out_dir;
            }
        }
        visual {
            CASAVAFilter {
                pos:"-810 -528";
                style:ext;
                bg-color-ext:"194 0 0 64";
                bounds:"-30 -30 181.875 92";
                in-file.angle:180;
                out-file.angle:360;
            }
            CutAdaptFastq {
                pos:"-452 -528";
                style:ext;
                bg-color-ext:"236 177 178 64";
                in-file.angle:180;
                out-file.angle:360;
            }
            MergeFastq {
                pos:"-107 -290";
                style:simple;
                bg-color-simple:"78 151 184 255";
                in-file.angle:90;
                out-file.angle:270;
            }
            QualityTrim {
                pos:"-144 -527";
                style:ext;
                bg-color-ext:"204 68 102 64";
                bounds:"-30 -30 106 96";
                in-file.angle:180;
                out-file.angle:290.925;
            }
            fastqc {
                pos:"-976 -294";
                style:simple;
                bg-color-simple:"0 128 128 255";
                in-file.angle:90;
            }
            fastqc-1 {
                pos:"-352 -254";
                style:simple;
                bg-color-simple:"0 128 128 255";
                in-file.angle:358.091;
            }
            get-file-list {
                pos:"-1086 -528";
                style:ext;
                bg-color-ext:"24 102 175 64";
                bounds:"-30 -30 102.25 92";
                out-url.angle:360;
            }
            tophat {
                pos:"-180 -119";
                style:ext;
                bg-color-ext:"0 128 0 64";
                in-sequence.angle:21.9745;
                out-assembly.angle:207.408;
            }
            CASAVAFilter.out-file->CutAdaptFastq.in-file {
                text-pos:"-38.75 -27";
            }
            CutAdaptFastq.out-file->QualityTrim.in-file {
                text-pos:"-40.75 -27";
            }
            MergeFastq.out-file->fastqc-1.in-file {
                text-pos:"-49.75 -28";
            }
            MergeFastq.out-file->tophat.in-sequence {
                text-pos:"-25.75 -19";
            }
            QualityTrim.out-file->MergeFastq.in-file {
                text-pos:"-48.75 -15";
            }
            get-file-list.out-url->CASAVAFilter.in-file {
                text-pos:"-39.7578 -27";
            }
            get-file-list.out-url->fastqc.in-file {
                text-pos:"-49.7578 25";
            }
        }
        wizard {
            name:"Raw RNA-Seq Data Processing Wizard";
            help-page-id:16122727;
            page {
                id:1;
                next:2;
                title:"Input data";
                parameters-area {
                    group {
                        title:"Sequencing reads";
                        label-size:120;
                        get-file-list.url-in {
                            label:"FASTQ files";
                        }
                    }
                }
            }
            page {
                id:2;
                next:3;
                title:Pre-processing;
                parameters-area {
                    group {
                        title:Filtration;
                        label-size:120;
                        QualityTrim.qual-id {
                        }
                        QualityTrim.len-id {
                            label:"Min length";
                        }
                        QualityTrim.both-ends {
                            label:"Trim both ends";
                        }
                        CutAdaptFastq.adapters-url {
                            label:Adapters;
                        }
                    }
                }
            }
            page {
                id:3;
                next:4;
                title:Mapping;
                parameters-area {
                    group {
                        title:"TopHat input";
                        label-size:180;
                        bowtie-index {
                            tophat.bowtie-index-dir {
                            }
                            tophat.bowtie-index-basename {
                            }
                        }
                        tophat.bowtie-version {
                        }
                    }
                    group {
                        title:Parameters;
                        label-size:170;
                        tophat.known-transcript {
                        }
                        tophat.raw-junctions {
                        }
                    }
                    group {
                        title:Additional;
                        label-size:170;
                        type:hideable;
                        tophat.mate-inner-distance {
                        }
                        tophat.mate-standard-deviation {
                        }
                        tophat.library-type {
                        }
                        tophat.no-novel-junctions {
                        }
                        tophat.max-multihits {
                        }
                        tophat.segment-length {
                        }
                        tophat.fusion-search {
                        }
                        tophat.transcriptome-only {
                        }
                        tophat.transcriptome-max-hits {
                        }
                        tophat.prefilter-multihits {
                        }
                        tophat.min-anchor-length {
                        }
                        tophat.splice-mismatches {
                        }
                        tophat.read-mismatches {
                        }
                        tophat.segment-mismatches {
                        }
                        tophat.solexa-1-3-quals {
                        }
                        tophat.bowtie-n-mode {
                        }
                        tophat.bowtie-tool-path {
                        }
                        tophat.samtools-tool-path {
                        }
                        tophat.path {
                        }
                        tophat.temp-dir {
                        }
                    }
                }
            }
            page {
                id:4;
                title:"Output data";
                parameters-area {
                    group {
                        title:"TopHat data";
                        label-size:130;
                        tophat.out-dir {
                            label:Tophat;
                        }
                    }
                    group {
                        title:"Filtered FASTQ";
                        label-size:130;
                        type:hideable;
                        CASAVAFilter.out-mode {
                        }
                        CASAVAFilter.custom-dir {
                        }
                    }
                }
            }
        }
    }
}