Hosting custom genomes or annotations

[manulera]

A possible way in which custom genomes or annotation could be hosted:

• Create a github repository where these genomes will be hosted
• Users can submit new genomes via a web form that creates a PR to the repository and runs some kind of processing pipeline, like in the submission repository.
• That stores the sequence as gb file and the annotation as gff
• In addition, from the gff it makes a json file with the same info to be queried from the frontend
• It also makes a single json file with the info of all the available genomes.

The good thing of this is that it's pretty straightforward, but it may be a bit slow depending on the network. The files should be loaded from the backend.

[manulera]

cc @dgruano

[dgruano]

I had a couple of thoughts about this:

• People may use reference genomes (not only new ones) with custom annotation. I'm thinking for instance people that want to extract sequences from genome browser tracks (like in PomBase, UCSC Browser etc...), so I would say we should be open to this in our implementation.

• Some of these custom annotations may be stored in reference databases. For instance, I am thinking:

  • Ensembl annotation of the human genome.
  • Lncipedia annotation of long non-coding RNAs
  • One of S. pombe's replication origin track

  While the two first have a proper version control and can be traced back, the latter is a different case (i.e. is a "single shot" annotation from a paper. Would adding them to the repo be our way to keep a trace of their origin? What happens if different users load the same annotation? It seems like we would be storing our own database of public genome annotations!

Happy to discuss further about this!