Genome and Annotation Idiosyncrasies

annotation

1. We are using the annotation file kfish2rae5g.main.pub.gff for many of our on-going analyses.
2. It has some issues.
3. Split=1, 2 etc, means that a gene model assembled from RNA-seq data does not map cleanly to the reference genome assembly and has been split into multiple GFF entries. These entries may be on different reference genome scaffolds, or they may be in different locations on the same scaffold. Split entries refer to the same transcript.
4. Some entries in the GFF have negative spans. E.g. Scaffold0 100 90. This throws an error in most any program that can read GFF. There aren't many of these entries, so we just throw them out.
5. There are some scaffold names in the GFF file (e.g. NOPATH and scaffold_1167) that do not occur in the reference genome assembly. For any application of the GFF to the reference genome these are useless and may result in errors (e.g. when using BEDTOOLs).
6. Transcript IDs, e.g. Funhe2EKm037517t1: Funhe2EKm is the assembly set. 037517 is an arbitrary ID number, but they tend to be in order, so 037517 is generally very close to 037518 in the assembly (though not always). t1 means this is the primary transcript for this gene. "Primary" does not mean that it is correct, complete, or even the most commonly expressed. It just means "primary".

genome

1. There are currently two assemblies

   1. Original assembly with the genome release paper.
      • fasta: killifish20130322asm.fa
      • this is what Noah used for the Science pollution adaptation paper
      • This assembly differs from the NCBI assembly, as of May 2017, in NCBI scaffold names relative to our scaffold names. NCBI deleted scaffold 6427, so everything beyond this scaffold is off by one. Talk to Andrew about the conversion table. Will be posted here at some point
   2. NCBI assemblies
      • Refseq version
      • Genbank version
      • Note that these two versions use different scaffold names. refseq begins NW_## while genbank is KN##
      • Look here for an explanation of the difference between the two versions.
      • Use the Refseq version, it has a good NCBI derived annotation
   3. We have also converted our annotation to work with the refseq and genbank genomes.
      • Link to gff files

2. Note that in fasta files scaffolds are not sorted by size.