quarto-revealjs-editable has a new release with 5.0.0!!
It is now much more stable, should now work everywhere, plus:
- Rotation support
- Undo/Redo
- Copy qmd to Clipboard
- keyboard controls
github.com/EmilHvitfeld...
#quarto
Issue 35 of #rdmweekly is out!
Including:
➡️ Building Realistic Fake Datasets with Pointblank @richmeister.bsky.social
➡️ Rethinking TXT Files @kbriney.bsky.social
➡️ Economic Benefits of OS @plos.org
➡️ What Is Data Governance? @thegovlab.org
and more!
Link: rdmweekly.substack.com/p/rdm-weekly...
📢 New preprint out!
tl;dr: Publish your code, add clear README's!
138 participants assessed data & code archiving practices across 1861 papers published at 7 @britishecologicalsociety.org journals, identified gaps & offered recommendations for improvement
🔗 doi.org/10.32942/X26...
"Dashboards" are very frustrating for exactly this reason. Needlessly fancy dashboards look impressive to people who don't use them.
OpenAI announces ChatGPT is now at the level of real scientific reviewers, “It just tells you to do single-cell RNAseq in response to every query” says Sam Altman
definitely a problem when dealing with diverse object structures. And you rely on the dev having generated (and documented!) those accessors. I feel like often there’s an "exotic" slot that was deemed unimportant so you have no choice but to @. S4 feels a bit like "don’t do this at home" to me 😅
The first time I got familiar with this is this epic rant from Aaron Lun on the Seurat R package to analyse single cell RNAseq data: github.com/LTLA/scRNAse...
S4 methods developers when you type @ to access some slots. #rstats
The closure who loved me
Have you looked at the data? Me:
I feel like we could collectively come up with some great bond flick names re: #rstats, #GitHub, etc 😂
I submit for your approval: terrible ideas only.
"MIT License to Kill"
"Rebase Another Day"
"The Dev Who Reviewed Me"
yes - there is certainly an uptick in fairly simple analyses. we've always had a policy excluding simple analyses (BLAST search etc) but what counts as simple is tricky to define and LLMs make a lot of things more simple to do... Others are seeing this too www.science.org/content/arti...
The dilemma is usually: do you want this to be a 2min or a 5 hours conversation?
A lot of computational biology relies on reusing publicly available data. I guess it’s probably easy to automatically generate a bunch of boiler plate analyses and write a manuscript with LLMs as well. Have you seen an uptick akin to this? Do you think it will be a significant problem?
Write R, ask Claude code to port it to the other language 🤪
Then I am biased towards tidyverse, but I think it’s nice to do "here is how you would do it in base R"
The way I was "brought up" in this: if you have no programming experience, you should not start with R from scratch but take a basic programming course first. I went through the CS50 course, did a bit of bash, python. And I think that was good background to have before diving into R.
I mean, it's completely fine to say there's a 95% chance that a tibble will not behave as a data frame.
(you can't deny that this was an epic example, incorporating baseR-tidyverse flamewar with frequentist-CI-interpretation-controversial)
Great to see this work out, led by Jon Rosen and co-supervised Karen Mohlke @klmohlke.bsky.social!
Connecting to some recent threads here:
I wrote something. I hope it can be the start of more conversations about AI and the choices we all have to make.
erincikanek.com/the-rumors-o...
A screenshot of two reply messages on bluesky commenting on the same post. The reply on top is very negative about the data.table package, saying 'it was just the worst'. The bottom reply instead claims that data.table's syntax is the best thing produced in #rstats.
Never change #rstats, never change
I just made an account and voted for this biology laboratory made of LEGO to "promote biological research and inspire more people into the world of biology". They need 10K votes. Reposting and/or voting below would be helpful. 🤗
beta.ideas.lego.com/product-idea...
This is (obviously) a step in the right direction: you should share all the code that led to the results (even if you think there is nothing "new" in your code). But who will review the code sharing?
> d3.js code with animations, but "vibe coding" scientific apps is likely to be very dangerous. I would need to completely gut this app of the backend stats functions and substitute my own.
