20221107 developer call notes

[hammer]

Pre-notes

• Notes from previous meeting

PRs

• tsinfer#748 Minimum viable sgkit dataset

Issues

• sgkit#947 potential issue in how ds.variant_allele is populated in read_plink
• sgkit#948 Improve release notes
• sgkit#924 VCF export
• Validation failure due to missing pooch lib (failure, need issue?)
• sgkit-publication#1 Identify target journals and article types
• sgkit-publication#2 Agree on outline

Notes

Attendees

1. Jeff Hammerbacher
2. Tom White
3. Jerome Kelleher
4. Ben Jeffery
5. Rafal Wojdyla
6. Alistair Miles

Discussion

• Tom’s contract w/ Oxford (until May): VCF and Plink writing
• tsinfer#748
  • Ready to merge!
  • Some challenges due to pins and lack of Windows support for cyvcf2
  • Follow up issues filed
  • Now need to work with ancestor data
  • TW: upper bound pins probably shouldn’t be done
  • A new sgkit release soon is probably a good idea
• sgkit#948
  • tskit: each PR that changes public behavior should include a changelog update
  • TW: who usually updates it? Contributors, unless it’s small and a new contributor
• sgkit#947
  • Issue with identifying major and minor alleles
  • Next step: write a test
  • TW will handle
• sgkit#924
  • TW initial benchmarks: 1 MB/s, 2 orders of magnitude off of our goal!
  • Working on INFO fields this week
• PC Relate validation failure due to missing pooch lib (failure, need issue?)
  • https://github.com/fatiando/pooch
  • TW: may have been a bed-reader release recently that caused it
  • RW: will create an issue (add upstream tag)
    • sgkit#949
• sgkit-publication#1
  • JH: Two years since we started talking about it. Landscape has changed a bit since then. Don’t want to just do an app paper.
  • JK: consider PeerJ
  • JH: Won’t get noticed there. Like a preprint server. Minimum get on PubMed. Put on bioRxiv.
  • JK: that’s fair. We want people to read the paper.
  • RW: PeerJ might be in PubMed?
  • JH: eLife? New model
  • JK: is what we are doing in its remit?
  • AM: F1000 research. Same platform as Welcome and Gates Open Research.
  • JH: Have used F1000 for posters. BMC journals were very slow in prev lab.
  • JH: Plink used gigascience for its 2.0 paper.
  • AM: could use malariagen as an example. Still mainly a software paper.
  • JK: we published stdpopsim paper in eLife.
  • JH: like to work with publisher we’ve used before.
  • JH: eLife paper types: ​​https://reviewer.elifesciences.org/author-guide/types
  • JK: sounds good.
  • AM: do you need to say which editor you want to work with?
  • JK: says authors should be prepared to suggest 2 editors.
  • JH: website says “Authors should be ready to suggest two of eLife's Senior Editors and at least two members of the Board of Reviewing Editors”. Find them at https://elifesciences.org/about/people. JK says Molly and Graham for reviewing editors. Ask Graham for suggestions of senior editors.
• sgkit-publication#2
  • JW: let’s brainstorm it out as a group!
  • JH: will ask Lorin Crawford if there’s a grad student to involve?
• JK: will try out sgkit for UKB 150k WGS
  • TW: upstream fix for Dask memory issues we should try out Share your experiences with `worker-saturation` config to reduce memory usage dask/distributed#7128
  • TW: read speed not being optimized right now