data.hjson

{
  data: [

    {
    name: Ebola Genome.fa
    value: export/local/bwa/AF086833.fa
    type: FASTA
    help: Ebola sequencing project.
    project_uid: cookbook
    }

    {
    name: Single End Reads.fq
    value: export/local/bwa/SRR1972739.fq
    type: FASTQ
    help: Single end sequencing reads.
    project_uid: cookbook
    }

    {
    name: FASTQ Data Collection
    value: export/local/fastqc/
    type: FASTQ
    help: This collections contains two different paired-end sequencing runs.
    project_uid: cookbook
    }

    {
    name: Fish Metabarcode Testdata
    value: export/local/usfish/data
    type: FASTQ
    help: This collections contains fish metabarcode data.
    project_uid: cookbook

    }

    {
    name: Samplesheet for Fish Metabarcode Testdata
    value: export/local/usfish/sample-sheet.csv
    type: CSV
    help: This file contains information on the metabarcode data.
    project_uid: cookbook

    }

    {
    name: Fish Sequences by Common Name
    value: export/local/usfish/fishname.fa
    type: FASTA
    help: Fish genomic sequences identified by common names.
    project_uid: cookbook
    }

    {
    name: Fish Sequences by Accession Number
    value: export/local/usfish/fishacc.fa
    type: FASTA
    help: Fish genomic sequences identified by accession numbers.
    project_uid: cookbook
    }


    {
    name: ADH partial region.fa
    value: export/local/augustus/adh.fa
    type: FASTA
    help: A 2.9 Mb long sequence from the Drosophila adh region (copied from the GASP dataset page)
    project_uid: cookbook
    }

    {
    name: ADH partial region.gff
    value: export/local/augustus/adh.gff
    type: GFF
    project_uid: cookbook
    help: Genes from a  2.9 Mb long sequence from the Drosophila adh region.
    }

    {
    name: RNA-Seq Paired End Collection
    value: export/local/rnaseq/reads-PE
    type: RNASEQ, FASTQ
    project_uid: cookbook
    help: This collection contains paired end data from 6 samples under 2 conditions(Griffith dataset subset).
    }

    {
    name: RNA-Seq Single End Collection
    value: export/local/rnaseq/reads-SE
    type: RNASEQ, FASTQ
    project_uid: cookbook
    help: This collection contains single end data from 6 samples under 2 conditions(Griffith dataset subset).
    }

    {
    name: Transcripts from chr22 human genome.fa
    value: export/local/rnaseq/refs/chr22_transcripts.fa
    type: FASTA, TRANSCRIPTS
    project_uid: cookbook
    help: Transcript sequences from human chromosome 22.
    }

    {
    name: Qiime2 testdata
    value: export/local/qiime2/data
    type: FASTQ, COLLECTION
    project_uid: cookbook
    help: Test dataset for qiime2 taxonomic classification.

    }
    {
    name: Test metadata sheet for Qiime2
    value: export/local/qiime2/12s_metadata_file_validated.csv
    type: CSV
    project_uid: cookbook
    help:Sample metadata for qiime2 classification testdata

    }

    {
    name: Reference file for Qiime2 testdata
    value: export/local/qiime2/12s-ref-qiime2.fasta
    type: FASTA
    project_uid: cookbook
    help:Reference sequence as fasta file.

    }

    {
    name: Samplesheet for PCR barcoded Testdata
    value: export/local/usfish/pcr-barcoded-samplesheet.csv
    type: CSV
    project_uid: cookbook
    help: This file contains information on the metabarcode data.

    }
    {
    name: Fish PCR Barcoded Testdata
    value: export/local/usfish/pcr-barcoded-data
    type: FASTQ
    project_uid: cookbook
    help: This collections contains fish pcr-barcoded data.

    }

    {
    name: Fish database.fa
    value: /export/data/usfish/reference/fishdb.fa
    type: FASTA
    project_uid:fish
    help: '''
    Fish specific DNA sequences .
    Provided as an example.
    '''
    }

    {
    name: Fish Metabarcode Testdata
    value: export/local/usfish/
    type: FASTQ, COLLECTION
    project_uid:fish
    help: This collections contains fish metabarcode data.

    }

    {
    name: Samplesheet for Fish Metabarcode Testdata
    value: export/local/usfish/sample-sheet.csv
    type: CSV
    project_uid:fish
    help: This file contains information on the metabarcode data.
    }


    {
    name: Sequencing Run 1
    value: /export/data/usfish/runs/run1
    type: FASTQ, COLLECTION
    project_uid:fish
    help: Data for sequencing run 1
    }

    {
    name: Sample sheet for Sequencing Run 1
    value: /export/data/usfish/runs/run1-sample-sheet.csv
    type: CSV
    project_uid:fish
    help: '''
    Specfications for the data in Sequencing Run 1
    This file is in a format agreed to when developing the recipe. It contains a number of columns
    that describe the samples names, barcodes, primers and file names.
    '''

    }

    {
    name: Sequencing Run 2
    value: /export/data/usfish/runs/run2
    type: FASTQ, COLLECTION
    project_uid:fish
    help: Data for sequencing run 2
    }

    {
    name: Sample sheet for Sequencing Run 2
    value: /export/data/usfish/runs/run2-sample-sheet.csv
    type: CSV
    project_uid:fish
    help: '''
    Sample description for Sequencing Run 2
    This file is in a format agreed to when developing the recipe. It contains a number of columns
    that describe the samples names, barcodes, primers and file names.
    '''

    }

    {
    name: Results: Genome Annotations.gtf
    value: /export/data/giraffe/results/annotations.gtf
    type: GTF, ANNOTATIONS
    project_uid: giraffe
    help: '''
    Annotations produced with Augustus and using RNA-Seq data.
    Contains annotations produced with Augustus using hints generated from the RNA-Seq data.
    '''
    }

    {
    name: Results: Protein Sequences.fa
    value: /export/data/giraffe/results/proteins.fa
    type: FASTA, PROTEIN
    project_uid: giraffe
    help: '''
    Predicted protein sequences.
    Contains the predicted protein sequences based on the annotations generated with Augustus.
    '''
    }

    {
    name: Results: Functional Annotations.txt
    value: /export/data/giraffe/results/functional-annotations.txt
    type: TXT
    project_uid: giraffe
    help: '''
    Functional annotations for each gene annotation.
    Contains the predicted transcript sequences annotated by their closest
    match to a known sequence.
    '''
    }

    {
    name: Assembly: Short Section.fa
    value: /export/data/giraffe/assembly/subset.fa
    type: FASTA
    project_uid: giraffe
    help: '''
    Contains the four largest contigs of the published assembly. Use it to test the pipeline.
    Contains the scaffolds scaffold320, scaffold268, scaffold1292,scaffold370 from the
    giraffe scaffold assembled by Rayan Chiki.
    '''

    }

    {
    name: Assembly: Published Giraffe Scaffold 2016.fa
    value: /export/data/giraffe/assembly/soap_assembly_k91_params.gapfilled.fa
    type: FASTA
    project_uid: giraffe
    help: '''
    The published assembled Giraffe scaffold (2016)
    The giraffe scaffold assembled by Rayan Chiki
    '''
    }

    {
    name: Sequencing Run: MA1_MP
    value: /export/data/giraffe/data/GIRAFFE/MA1_MP
    type: FASTQ
    project_uid: giraffe
    help: '''
    Giraffe, MA1, Mate Pairs, 150bp, 2 x 78 million reads, 8x coverage
    Specimen ID: `MA1`

    Species: Masai Giraffe

    Sample: Female, Masai Giraffe liver,  Kenya

    Mate Paired Library

    Date created: 2014-Jan-14

    Date submitted: 2016-Mar-14

    SRA ID: `SRR3217431`

    Stats:

    ```
    adultfemalegiraffeMA1_R1.fastq.gz  FASTQ   DNA   78,494,994  11,774,249,100      150      150      150
    adultfemalegiraffeMA1_R2.fastq.gz  FASTQ   DNA   78,494,994  11,774,249,100      150      150      150
    ```

    '''
    }

    {
    name: Sequencing Run: MA1_PE
    value: /export/data/giraffe/data/GIRAFFE/MA1_PE
    type: FASTQ
    project_uid: giraffe
    help: '''
    Giraffe, MA1,  Paired End, 150bp, 2 x 285 million reads, 28x coverage
    Specimen ID: `MA1`

    Species: Masai Giraffe

    Sample: Female, Masai Giraffe liver,  Kenya

    DNA-Sequencing, Paired End Library, average coverage 28x

    Date created: 2013-Jul-26

    Date submitted: 2016-Mar-14

    SRA ID: `SRR3218456`

    Stats:

    ```
    MA1_R1.fastq.gz  FASTQ   DNA   285,111,308  42,766,696,200      150      150      150
    MA1_R2.fastq.gz  FASTQ   DNA   285,111,308  42,766,696,200      150      150      150
    ```

    '''
    }

    {
    name: Sequencing Run: NZOO_PE
    value: /export/data/giraffe/data/GIRAFFE/NZOO
    type: FASTQ
    project_uid: giraffe
    help: '''
    Giraffe, NZOO, Paired End, 151bp, 4 x 22 million reads, 30x coverage
    Specimen ID: `NZOO`

    Species: Masai Giraffe

    Sample: Female, Masai Giraffe liver, Nashville Zoo

    DNA-Sequencing, Paired End, 151bp length

    Stats:

    ```
    nzoo_L001_r1.fq.gz  FASTQ   DNA   151,156,603  22,824,647,053      151      151      151
    nzoo_L001_r2.fq.gz  FASTQ   DNA   151,156,603  22,824,647,053      151      151      151
    nzoo_L002_r1.fq.gz  FASTQ   DNA   152,207,809  22,983,379,159      151      151      151
    nzoo_L002_r2.fq.gz  FASTQ   DNA   152,207,809  22,983,379,159      151      151      151
    ```

    '''
    }

    {
    name: Sequencing Run: UGA_PE
    value: /export/data/giraffe/data/GIRAFFE/UGA
    type: FASTQ
    project_uid: giraffe
    help: '''
    Giraffe, NZOO, Paired End, 250bp, 2 x 333 million reads, 55x coverage
    Specimen ID: `UGA`

    Species: Rothschild Giraffe

    Sample: Fibroblast Cell Cultures, San Diego Zoo Institute for Conservation Research

    DNA-Sequencing, Paired End Library, 250bp long, 2 x 26 million reads

    Expected coverage: 22x

    Stats:

    ```
    UGA_R1.fastq.gz  FASTQ   DNA   333,671,563  83,417,890,750      250      250      250
    UGA_R2.fastq.gz  FASTQ   DNA   333,671,563  83,417,890,750      250      250      250
    ```

    '''
    }

    {
    name: Sequencing Run: WOAK_PE
    value: /export/data/giraffe/data/OKAPI/WOAK_PE
    type: FASTQ
    project_uid: giraffe
    help: '''
    Okapi, WOAK_PE, Paired End, 150bp, 4 x 135 million reads, 27x coverage
    Specimen ID: `WOAK`

    Species: Okapi

    Sample: Fetal Male Okapi Liver, White Oak Holdings

    DNA-Sequencing, Paired End Library, 150bp read lenghs

    Stats:

    ```
    lane1_R1.fastq.gz  FASTQ   DNA   133,895,515  20,084,327,250      150      150      150
    lane1_R2.fastq.gz  FASTQ   DNA   133,895,515  20,084,327,250      150      150      150
    lane2_R1.fastq.gz  FASTQ   DNA   136,707,167  20,506,075,050      150      150      150
    lane2_R2.fastq.gz  FASTQ   DNA   136,707,167  20,506,075,050      150      150      150
    ```

    '''
    }

    {
    name: Sequencing Run: WOAK_MP
    value: /export/data/giraffe/data/OKAPI/WOAK_MP
    type: FASTQ
    project_uid: giraffe
    help: '''
    Okapi, WOAK_MP, DNA, Mate Pair,  150bp, 2 x 67 million reads, 7x coverage
    Specimen ID: `WOAK`

    Species: Okapi

    Sample: Fetal Male Okapi Liver, White Oak Holdings

    DNA-Sequencing, Mate Pair, 150bp, 2 x 67 million reads, 7x coverage

    Stats:

    ```
    fetalokapi1_R1.fastq.gz  FASTQ   DNA   67,300,363  10,095,054,450      150      150      150
    fetalokapi1_R2.fastq.gz  FASTQ   DNA   67,300,363  10,095,054,450      150      150      150
    ```

    '''
    }

    {
    name: Sequencing Run: NZOO_RNASEQ,
    value: /export/data/giraffe/data/GIRAFFE/RNASEQ
    type: FASTQ, RNASEQ
    project_uid: giraffe
    help: '''
    Giraffe, NZOO, RNASEQ, 150bp,  4 x 50 million, for Liver, Heart, Lung and Fibroblast Cell Line
    Specimen ID: `NZOO` and 'MASG'
    ```
    GiraffeCellLine_R1.fastq.gz  FASTQ   DNA   50,073,121  7,510,968,150      150      150      150
    GiraffeHeart_R1.fastq.gz     FASTQ   DNA   40,082,453  6,012,367,950      150      150      150
    GiraffeLiver_R1.fastq.gz     FASTQ   DNA   49,715,450  7,457,317,500      150      150      150
    GiraffeLung_R1.fastq.gz      FASTQ   DNA   54,637,776  8,195,666,400      150      150      150
    ```
    '''
    }

    {
    name: Sequencing Run: WOAK_RNASEQ
    value: /export/data/giraffe/data/OKAPI/RNASEQ
    type: FASTQ, RNASEQ
    project_uid: giraffe
    help: '''
    Okapi, WOAK, OKSD, RNASEQ, 150bp, 2 x 56 million, Liver and Fibroblast Cell Line
    Specimen ID: `WOAK` and `OKSD`

    ```
    OkapiCellLine_R1.fastq.gz  FASTQ   DNA   56,933,092  8,539,963,800      150      150      150
    OkapiLiver_R1.fastq.gz     FASTQ   DNA   57,133,111  8,569,966,650      150      150      150
    ```
    '''
    }

    {
    name: Transcripts from chr22 human genome.fa
    value: export/local/rnaseq/refs/22.fa
    type: FASTA, TRANSCRIPTS
    project_uid:handbook
    help: Transcript sequences for chr22 of the human genome.
    }

    {
    name: Gene Annotations for chr22 human genome.gtf
    value: export/local/rnaseq/refs/22.gtf
    type: GTF
    project_uid:handbook
    help: Transcript and gene models for chr22 of the human genome.
    }

    {
    name: ERCC92 Spike Control Reference.fa
    value: export/local/rnaseq/refs/ERCC92.fa
    type : FASTA, TRANSCRIPTS
    project_uid:handbook
    help: Transcript sequences for the ERCC92 Spike Control.
    }

    {
    name: ERCC92 Spike Control Annotations.gtf
    value: export/local/rnaseq/refs/ERCC92.gtf
    type : GTF
    project_uid:handbook
    help: Transcript and gene models for the ERCC92 Spike Control .
    }

    {
    name: Metagenomics Data Directory
    value: /export/data/mothur-study1
    type: MOTHUR
    project_uid: mothur
    help: '''
    A prepackaged data set with test files.
    Contains reads in a directory and a file list.
    '''
    }

    {
    name: Fish Sequences by Accession Number
    value: export/local/usfish/fishacc.fa
    type: FASTA
    project_uid:trout
    help: Fish genomic sequences identified by accession numbers.
    }

    {
    name: Fish Sequences by Common Name
    value: export/local/usfish/fishname.fa
    type: FASTA
    project_uid:trout
    help: Fish genomic sequences identified by common name.
    }

    {
    name: Fish Metabarcode Testdata
    value: export/local/usfish/
    type: FASTQ, COLLECTION
    project_uid:trout
    help: This collections contains fish metabarcode test data.

    }

    {
    name: Samplesheet for Fish Metabarcode Testdata
    value: export/local/usfish/sample-sheet.csv
    type: CSV
    project_uid:trout
    help: This file contains information on the metabarcode data.
    }

    {
    name: Sequencing Run 1
    value: /export/data/usfish/runs/run1
    type: FASTQ, COLLECTION
    project_uid:trout
    help: Data for sequencing run 1
    }

    {
    name: Sample sheet for Sequencing Run 1
    value: /export/data/usfish/runs/run1-sample-sheet.csv
    type: CSV
    project_uid:trout
    help: '''
    Specfications for the data in Sequencing Run 1
    This file is in a format agreed to when developing the recipe. It contains a number of columns
    that describe the samples names, barcodes, primers and file names.
    '''

    }

    {
    name: Sequencing Run 2
    value: /export/data/usfish/runs/run2
    type: FASTQ, COLLECTION
    project_uid:trout
    help: Data for sequencing run 2
    }

    {
    name: Sample sheet for Sequencing Run 2
    value: /export/data/usfish/runs/run2-sample-sheet.csv
    type: CSV
    project_uid:trout
    help: '''
    Sample description for Sequencing Run 2
    This file is in a format agreed to when developing the recipe. It contains a number of columns
    that describe the samples names, barcodes, primers and file names.
    '''
    }

    {
    name: Genome.fa
    value: recipes/tutorial/data/genome.fa
    type: FASTA
    project_uid:tutorial
    help: '''
    A test genome file.
    A longer description for the test sequence file goes here.
    '''
    }

    {
    name: SRR numbers.txt
    value: recipes/tutorial/data/sra.txt
    type: SRR
    project_uid:tutorial
    help: '''
    SRR sequencing run accession numbers from SRA.
    Each line represents a SRR run id.
    '''
    }

    {
    name: File Collection
    value: recipes/tutorial/collection
    project_uid:tutorial
    help: '''
    Multiple files and folders may be represented as a single data entry.
    This collection contains multiple files and folders. The data view lists the content of the directory.
    '''
    }


]
}