large sequences

[phiweger]

We are likely to walk into genomes too large to efficiently store in MongoDB w/ its max doc size of 16 mb.

• GridFS?
• supplementary fasta with the primary document key as header

I prefer the latter, because it seems less dependent on the db architecture.

We could implement random access to fasta (or use pyfaidx) easily, so would be quite efficient to retrieve a given sequence once we're computing stuff.

[phiweger]

if seq empty:

• link: accession/ filepath, UUID from _id field

links to related sequences

• link: accession/ filepath, someid, description