---
directories:
    work_dir: /usr/local/www/data/private/Arachis/hypogaea/Tifrunner.gnm2.ann2
    from_annot_dir: adjud4
    from_genome_dir: Tifrunner.gnm2
prefixes:
    from_annot_prefix: "arahy.Tifrunner.gnm2.ann2." 
    from_genome_prefix: "arahy.Tifrunner.gnm2."
collection_info:
    genus: Arachis
    species: hypogaea
    scientific_name_abbrev: arahy
    coll_genotype: Tifrunner
    gnm_ver: gnm2
    ann_ver: ann2
    genome_key: J5K5
    annot_key: PVFB
readme_info:
    provenance: "The files in this directory are a product of the International Peanut Genome Initiative. The files here are held as part of the LegumeInfo and Peanutbase 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 Peanutbase 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://data.legumeinfo.org"
    synopsis_genome: Genome assembly 2 for Arachis hypogaea, cultivar Tifrunner
    synopsis_annot: Annotation 2, using the BIND pipeline, for Arachis hypogaea, cultivar Tifrunner
    taxid: "3818"
    genotype: Tifrunner
    description: "Genome assembly for Arachis hypogaea cultivar Tifrunner. This version is essentially the same as Tifrunner version 1 (arahy.Tifrunner.gnm1.KYV3.genome_main.fna) at the basepair level, but has several large structural rearrangements, on the basis of HiC assessments. The largest changes are: a translocation of ~8 Mbp in the pericentromere of A02; several rearrangements in the pericentromere of A03; a swap in the telomeric regions of A06 and B06; inversion of a small scaffold from the distal end of A08; inversion of a telomeric region in B05; other small changes in telomeric regions of several pseudomolecules: B10, B09, A08."
    chromosome_prefix: chr
    supercontig_prefix: scaffold
    description_annot: "This annotation was produced by Jacqueline Campbell, Connor Cameron, and Andrew Farmer, using the BIND annotation pipeline."
    bioproject: "PRJNA419393"
    sraproject: 
    dataset_doi_genome: 
    dataset_doi_annot: 
    genbank_accession: 
    original_file_creation_date: "2023-03-31"
    local_file_creation_date: "2023-03-31"
    dataset_release_date: "2023-03-31"
    contributors: "The International Peanut Genome Initiative; lead assembly group Jeremy Schmutz, Jerry Jenkins, Jane Grimwood; project leads David Bertioli; Soraya Bertioli; Brian Schleffler; Scott Jackson; Peggy Ozias-Akins. Annotation by Jacqueline Campbell, Connor Cameron, and Andrew Farmer."
    publication_doi: "10.1038/s41588-019-0405-z"
    citation: "Bertioli, D.J., Jenkins, J., Clevenger, J. et al. The genome sequence of segmental allotetraploid peanut Arachis hypogaea. Nat Genet 51, 877-884 (2019). https://doi.org/10.1038/s41588-019-0405-z"
    publication_title: "The genome sequence of segmental allotetraploid peanut Arachis hypogaea"
    data_curators: Steven Cannon, Andrew Farmer
    public_access_level: public
    license: open
    keywords: "peanut, Arachis hypogaea, Tifrunner"
from_to_genome:
  - 
    from: genome_main_bare.fna.gz
    to: genome_main.fna
    description: "Primary genome assembly"
original_readme_and_usage:
from_to_annot_as_is:
from_to_genome_as_is:
from_to_cds_mrna:
  - 
    from: cds.fna.gz
    to: cds.fna
    description: "cds sequences"
  - 
    from: cds_lowqual.fna.gz
    to: cds_lowqual.fna
    description: "cds sequences lacking homology to Arachis gene models from other annotations"
  - 
    from: mrna.fna.gz
    to: mrna.fna
    description: "mRNA sequences"
  - 
    from: mrna_lowqual.fna.gz
    to: mrna_lowqual.fna
    description: "mRNA sequences lacking homology to Arachis gene models from other annotations"
from_to_protein:
  - 
    from: protein.faa.gz
    to: protein.faa
    description: "Protein sequences"
  - 
    from: protein_lowqual.faa.gz
    to: protein_lowqual.faa
    description: "Protein sequences lacking homology to Arachis gene models from other annotations"

from_to_gff:
  - 
    from: gene_models_bare.gff3.gz
    to: gene_models.gff3
    description: "Gene models - all (temporary; will be split into gene_models_main and gene_models_lowqual in post-processing)"

#from_to_gff:
#  - 
#    from: gene_models_main.gff3.gz
#    to: gene_models_main.gff3
#    description: "Gene models - main"
#  - 
#    from: gene_models_lowqual.gff3.gz
#    to: gene_models_lowqual.gff3
#    description: "Gene models - lacking homology to Arachis gene models from other annotations"