State of BioJulia packages (August 2023)

[M-PERSIC]

Packages that harmonize well: BBI (v0.2), BED (v0.3), BGZFStreams (v0.3), BigWig (v0.1), BioAlignments (v3), BioFetch (v0.2), BioGenerics (v0.1), BioSequences (v3), BioStructures (v2), Codecs (v0.5), FASTX (v2), FormatSpecimens (v1), GFF3 (v0.2), GenomicAnnotations (v0.3), GenomicFeatures (v2), IndexableBitVectors (v1), Indexes (v0.1), IntervalTrees (v1), Kmers (v0.1), Libz (v1), MMTF (v1), PopGen (v0.6), SequenceVariation (v0.2), SubstitutionModels (v0.5), Xam (v0.3)

NOTE: Automa is a hard dependency for many of these packages and because v1.0 was only recently introduced, I had to set the compatibility bound to v0.8 for the moment.

Packages that do not harmonize:

• [ ] SingleCellProjections
  • Requires Julia 1.8+
• [ ] ReadDatastores
  • Compatibility constraints
• [ ] GFF3
  • Automa compatibility constraint
• [ ] XAM
  • Automa compatibility constraint
• [ ] PopGenCore
  • Precompile failure
• [ ] WaveletMatrices
  • Not available in the public registry
• [ ] KmerAnalysisMakie
  • BioSequences compatibility unsatisfied
• [ ] KmerAnalysis
  • BioSequences compatibility unsatisfied
• [ ] GeneticVariation
  • Low compatibility bounds (Automa, BioSequences,...)
• [ ] GraphicalFragmentAssembly
  • Not available in the public registry
• [ ] BioTools
  • BioSequences compatibility constraint
• [ ] FMIndexes
  • Not available in the public registry
• [ ] Tabix
  • Empty :(
• [ ] GenomeGraphs
  • BioSequences, FASTX compatibility constraints
• [ ] BigBed
  • Automa compatibility constraint
• [ ] WAHVectors
  • Not available in the public registry
• [ ] FileIO
  • Is a fork and it's way behind
• [ ] BSG
  • Empty :(
• [ ] GenomeBrowsers
  • Not available in the public registry
• [ ] Phylogenies
  • Not available in the public registry
• [ ] NaturalSelection
  • Not available in the public registry
• [ ] CountMin
  • Is a fork and it's way behind
• [ ] juju
  • Empty :(
• [ ] fogsaa
  • Is a fork and has been unmaintained since 2016

[kescobo]

I'm not sure that we want to include literally everything. I think BioSequences, the file loaders (eg XAM, FASTX), Fetch, Tools, Alignments, Structures would be a good start.

Also, I'm not 100% sure that having this as an actual package is necessary, though I think this is a super useful exercise to gauge compatibility across the org and identify things that need to be shelved.

[kescobo]

Some additional thoughts on this list

1. I don't think any of the forks are necessary to hang onto - I think Sabrina was forking into the org instead of her personal account for some attempts to build compatibility, but AFAIK we're not trying to maintain any forks separate from the trunk
2. juju can definitely be deleted. It was never developed and in any case is obviated by juliaup
3. We should definitely open issues in things that are limited by Automa or BioSequences compat

We should probably add RepoStatus badges to everything that's not being actively maintained by someone (and the others too, probably)

[M-PERSIC]

I agree. The good news is that the only true blockers for almost every package being included are the compat bounds and Automa. I made this issue so that the community could discuss each of these together :)

I don't think this should be a public package, but as a testbed for the ecosystem I think it should be considered. It's more for convenience that anything and can help with benchmarking and other purposes like with the CI workflow idea.

[kescobo]

Cool. Separately, we should add downstream tests to Automa and BioSequences.