Support FASTA format

[sehilyi]

https://github.com/higlass/higlass-sequence

[etowahadams]

Currently sequences data can be used in gosling through data "type": "multivec" . An example of this is in multiscale sequence track. It shows the sequence as a track with "mark": "bar" and overlays the letters below a certain zoom threshold.

We also want to be able to import data as fasta files. The gosling schema should accept a "type": "fasta" The HiGlass sequence viewer also requires that the fasta file be indexed and the chromosome sizes be specified. See below for an example from HiGlass:

 "data": {
   "type": "fasta",
   "fastaUrl": "https://aveit.s3.amazonaws.com/higlass/data/sequence/hg38.fa",
   "faiUrl": "https://aveit.s3.amazonaws.com/higlass/data/sequence/hg38.fa.fai",
   "chromSizesUrl": "https://aveit.s3.amazonaws.com/higlass/data/sequence/hg38.mod.chrom.sizes"
 },

Decisions to make:

1. Should we always require the user to index fasta files? Previous discussions on gff and vcf suggest this may be the path we should go down.
2. Should we have a dedicated sequence viewer track the way that HiGlass does? Or should we continue to have "mark": "bar" with overlaid text be the main way that sequences are viewed?

[sehilyi]

Should we always require the user to index fasta files? #764 (comment) on gff and vcf suggest this may be the path we should go down.

I think we need to make an index file required -- I am unsure if it would be possible to support smooth zooming/panning w/o it. The required properties can be like the following

{ // type of FastaData
  type: 'fasta',
  url: string,
  indexUrl: string
}

FYI, we do not use HiGlass' chromSizesUrl. Instead we let users define assembly.

Should we have a dedicated sequence viewer track the way that HiGlass does? Or should we continue to have "mark": "bar" with overlaid text be the main way that sequences are viewed?

I think we want to let users choose the visual mark that they wish to use, although I believe bar/text will be the typical mark types for such data.

[sehilyi]

For the second question, we could implement a track template for fasta data, e.g., template: 'fasta', to make the construction of a fasta-based sequence track easier. You can find some examples in our Editor (the very last example in the gallery).