Kelly Sovacool

@kelly.sova.cool

Bioinformatics Software Engineer. PhD from Schloss Lab at UMich. https://sovacool.dev 🏃🏻‍♀️🚴🏻‍♀️🧗🏻‍♀️ #python #rstats #nextflow #snakemake she/her My views are my own.

i started this train of thought as a joke but now i do think there's something to the idea that the way to sell the green revolution to the US public is that we're being denied Good Deals on cheap chinese cars and solar panels and there's nothing more unamerican than saying we can't have a good deal

Let's learn a lesson from the nearly 500K of us who mobilized to comment on the OMB rules. We can leave the sidelines, take action and it can make a difference. After 15 months of inaction, our scientific societies finally woke up and mobilized members. It doesn't stop there 1/n

496,769 comments received

I'm presenting poster C-049 at #ISMB2026 #BOSC2026 on Wednesday: "Anatomy of a modular toolchain for reproducible bioinformatics workflows" (Poster C-049). Please reach out if you're here and would like to meet up, and stop by my poster if you want to talk about reproducible software!

The anatomy of our modular toolchain for reproducible bioinformatics workflows. The central component is a Bioinformatics Workflow or Pipeline repository. Lines extend out of the Workflow to the other components: a Python package containing helper functions for a custom CLI, reusable Nextflow modules and sub workflows, Docker recipes, Documentation website, and Continuous integration. Each of these components contribute to scalability, reusability, and reproducibility of bioinformatics pipelines.

When I explain to peeps how much we do in research labs for how few $$, they simply don’t believe me. I once explained it to a famous person from Pixar, he looked me square in the eye and said “You mean all the cancer, vision, and Alzheimer’s grants work that way? You’ve gotta be %#ing kidding me!”

Jason Rasgon@vectorgen.bsky.social · last mo.

The NIH is an ENORMOUS bargain for the American people. Do folk think that private industry will do the background slog work for drug discovery for 250K direct costs?

Currently listening to a playlist that so far has included Herb Alpert, Tame Impala, Star Wars soundtrack (Duel of the Fates), and now Lil Nas X. Confused by the range but not upset about it!

In the absence of aggressive quality control it's just the mass production of tech debt that will inevitably collapse. I fear we'll eventually hit the finding out phase 😬

I have a similar experience. I am able to get the #Rstats code I want from Claude in just a few prompts, often in just one. It's not just that Claude has improved, it's also that I started using R in 2014 (I know what to ask) and I learned how to prompt effectively (I know how to ask).

Jordan Nafa@ajordannafa.com · 3mo ago

The more I use AI for development, the more concerned I am about the fact that knowing what makes an effective prompt and the appropriate, opiniated, guardrails necessary to consistently get high quality output is much harder if you don't have quite a bit of experience doing things by hand first.

The discourse around "This is a tell for AI writing" really only serves the AI bros, because (they believe) it will tell them how to change the AI to make it indistinguishable from a human. We need to keep our em dashes. Our rules of three. The tell for AI is: is it hollow, does it say nothing.

Sunny Moraine@dynamicsymmetry.bsky.social · 3mo ago

Why should I stop using em dashes? Why should I stop listing things in threes? They’re the ones who suck