---
directories:
    work_dir: /usr/local/www/data/private/Arabidopsis/thaliana
    from_annot_dir: Athaliana_447_Araport11/annotation
    from_genome_dir: Athaliana_447_Araport11/assembly
prefixes:
    from_annot_prefix: "Athaliana_447_Araport11."
    from_genome_prefix: "Athaliana_447_TAIR10."
collection_info:
    genus: Arabidopsis
    species: thaliana
    scientific_name_abbrev: arath
    coll_genotype: Col0
    gnm_ver: gnm9
    ann_ver: ann11
    genome_key: YF2W
    annot_key: KH24
readme_info:
    provenance: "The files in this directory originated from http://phytozome.jgi.doe.gov, but originated in the TAIR and Araport projects. Files in this present directory are derived, and may have changes, as noted below. The files here are held as part of the LegumeInfo and SoyBase projects, and are made available here for the purpose of reproducibility of analyses at these sites (e.g. gene family alignments and phylogenies, genome browsers, etc.) and for further use by researchers, as that research extends other analyses at the LegumeInfo and SoyBase projects. If you are conducting research on large-scale data sets for this species, please consider retrieving the data from the primary repositories. If you use the data in the present directory, please respect any usage restrictions in the present and original repositories, and cite the data appropriately."
    source: "https://phytozome-next.jgi.doe.gov/info/Athaliana_Araport11"
    synopsis_genome: Arabidopsis thaliana TAIR9 genome assembly
    synopsis_annot: Araport11 annotation on the TAIR9 genome assembly
    taxid: "3702"
    genotype: Col0
    description_genome: "This is the TAIR9 genome assembly of the Col-0 ecotype, produced by the Arabidopsis Genome Initiative."
    chromosome_prefix: chr
    supercontig_prefix: scaffold
    description_annot: "This is the Araport11 annotation, released in 2016, on the TAIR9 genome assembly of the Col-0 ecotype."
    bioproject: 
    sraproject: 
    dataset_doi_genome: 
    dataset_doi_annot: 
    genbank_accession: 
    original_file_creation_date: 2016-11-08
    local_file_creation_date: 2022-11-27
    dataset_release_date: 2022-11-27
    contributors: The Arabidopsis Genome Initiative, TAIR, and Araport projects
    publication_doi: "10.1111/tpj.13415"
    citation: "Cheng CY, Krishnakumar V, Chan AP, Thibaud-Nissen F, Schobel S, Town CD. Araport11: a complete reannotation of the Arabidopsis thaliana reference genome. Plant J. 2017 Feb;89(4):789-804. doi: 10.1111/tpj.13415. Epub 2017 Feb 10. PMID: 27862469."
    publication_title: "Araport11: a complete reannotation of the Arabidopsis thaliana reference genome."
    data_curators: Steven Cannon
    public_access_level: public
    license: open
    keywords: "Arabidopsis, Col-0, Col0, Columbia-0, Araport, TAIR"
from_to_genome:
  - 
    from: fa.gz
    to: genome_main.fna
    description: "Primary genome assembly"
original_readme_and_usage:
  - 
    from_full_filename: Athaliana_447_Araport11.readme.txt
    to: original_readme.txt
    description: "Original Araport11 data README file"
from_to_annot_as_is:
  - from: annotation_info.txt
    to: annotation_info.txt
    description: "Table of gene annotations from Araport"
  - from: defline.txt
    to: defline.txt
    description: "Defline for genes from Araport"
  - from: synonym.txt
    to: synonym.txt
    description: "Gene identifier synonyms, gene symbols, and functional descriptors"
from_to_genome_as_is:
from_to_cds_mrna:
  - 
    from: cds.fa.gz
    to: cds.fna
    description: "cds sequences"
  - 
    from: cds_primaryTranscriptOnly.fa.gz
    to: cds_primary.fna
    description: "cds sequences - primary only"
  - 
    from: transcript.fa.gz
    to: mrna.fna
    description: "Transcript sequences"
  - 
    from: transcript_primaryTranscriptOnly.fa.gz
    to: mrna_primary.fna
    description: "Transcript sequences - primary only"
from_to_protein:
  - 
    from: protein_primaryTranscriptOnly.fa.gz
    to: protein_primary.faa
    description: "Protein sequences - primary only"
  - 
    from: protein.fa.gz
    to: protein.faa
    description: "Protein sequences"
from_to_gff:
  - 
    from: gene_exons.no_suffix.gff3.gz
    to: gene_models_exons.gff3
    description: "Gene models, with exon features"
  - 
    from: gene.no_suffix.gff3.gz
    to: gene_models_main.gff3
    description: "Gene models - main"