iiinova — where your digital self keeps evolving

Sabrina Nystrom @handle.invalid · 28/09/2026

@stdlib.bsky.social's jj config has some real gems like 'sandwich' that I use constantly with the megamerge flow. I've cribbed a lot from it.

020

Sabrina Nystrom @handle.invalid · 01/06/2026

CyO flies run faster than normal! No free lunch.

020

Sabrina Nystrom @handle.invalid · 28/05/2026

I wish there was a way to know you're in the good old days before you've actually left them.

020

Sabrina Nystrom @handle.invalid · 28/05/2026

I miss getting issue reports that were just like "fuck you your software is broken please fix it now I need it urgently to work"

050

Sabrina Nystrom @handle.invalid · 28/05/2026

That's a great point Josiah, you're right to push back

140

Sabrina Nystrom @handle.invalid · 29/01/2026

DO NOT LOOK AT THE SOURCE CODE DO NOT LOOK AT THE SOURCE CODE DO NOT LOOK AT THE SOURCE CODE i may never recover

120

Sabrina Nystrom @handle.invalid · 17/11/2025

Tough to say without more details, but: provide an argument or an R option that allows users to enable the new behavior, but default to the old style while returning a warning that the old behavior will be deprecated in a later version. Later, alter the default before removing the old one.

010

Sabrina Nystrom @handle.invalid · 15/08/2025

Pick a time, I'll be there!

000

Sabrina Nystrom @handle.invalid · 15/08/2025

You can pry emacs from my cold dead fingers. That being said, if you're not using this, you're missing out: github.com/kahole/edama...

github.com

GitHub - kahole/edamagit: Magit for VSCode

Magit for VSCode. Contribute to kahole/edamagit development by creating an account on GitHub.

130

Sabrina Nystrom @handle.invalid · 03/12/2024

I love eval(parse()) crimes. I set the additional challenge that I wanted it to be a single string that got eval'd & only using base regex methods.

A screenshot of R code solutions to Part 1 and Part 2 of the Advent of code puzzle. There are way too many regular expressions.

120

Sabrina Nystrom @handle.invalid · 03/12/2024

I was just messing with this. No code written, but my plan was: `uv export --format requirements-txt`, sed away the versions, then `uv add -r requirements.txt`.

010

Sabrina Nystrom @handle.invalid · 13/11/2024

But have you considered the career advantages of discovering a new -ome that only you uniquely have the magic touch to investigate

000

Sabrina Nystrom @handle.invalid · 25/03/2024

Funny enough, this exists here: bsky.app/profile/xblo...

010

Sabrina Nystrom @handle.invalid · 29/09/2023

If I'm paying for software, I want the method to be great, yes, but what we really *want* to be buying is stability and maintenance so we can focus on our core products.

021

Sabrina Nystrom @handle.invalid · 29/09/2023

It was really disappointing. Ultimately came down to: inflexible legal / payment agreements (in one case wanting a several year commitment & full payment up front), and there's no guarantee the project won't die anyway due to funding lapses, etc. in which case we lose access, so becomes too risky.

110

Sabrina Nystrom @handle.invalid · 29/09/2023

Been on the purchasing side a time or two and we just decided to give up rather than deal with the headache.

110

Sabrina Nystrom @handle.invalid · 26/09/2023

Could also run something like mFOLD but I'm not fully convinced that translates to efficacy, but at the very least might indicate which ones are easier/harder to clone.

010

Sabrina Nystrom @handle.invalid · 26/09/2023

At that point I'd dip into genetic variants. More consistent seq across population == better? Avoid runs of same nt, prefer terminal GC clamp (or other platform-specific discriminators)? I usually make a matrix of these scores & minimize (can weight or not) No clue about the alt-nt pairing algo.

110

Sabrina Nystrom @handle.invalid · 26/09/2023

If the list is small, order em all?

100

Sabrina Nystrom @handle.invalid · 26/09/2023

(Where k is the length of your construct)

100

Sabrina Nystrom @handle.invalid · 26/09/2023

For ASOs I'm not as sure (other considerations there for alt-nt's for stability, etc), but for RNAi, it can be hard to predict what works best. If you have a fusion or something you can try spanning the boundary. I like to grab all kmers in ROI & subset to unique ones vs the masked genome.

100

Sabrina Nystrom @handle.invalid · 26/09/2023

We are now well beyond my realm of expertise!

110

Sabrina Nystrom @handle.invalid · 26/09/2023

I guess it depends most on your delivery vehicle. Nusinersen works so well (in part) cause you can just inject directly into the spine. Not sure your cancer type of interest, but I think that is key here if the end goal is 'quickest to patient'. If you have to go through the liver, 🤷‍♀️

110

Sabrina Nystrom @handle.invalid · 25/09/2023

Ok but Fiesta Grill is at least an 11/10, I'd hardly call that low standards.

120

Sabrina Nystrom @handle.invalid · 25/09/2023

Are you more interested in knockdown, or specifically using ASOs to do the knockdown? Admittedly not familiar with implementing ASOs in the lab, but shRNAs are pretty easy to design & test a lot of.

110