That was the original justification, but I can imagine them continuing because it saves them money and sub-30th percentiles are unlikely to get funded anyway. The large shift this year, plus the grants being resubmitted that would normally funded, seems likely to lead to score compression up top.
Gotcha, that makes sense.
From looking at past distributions, regular NDs seem to get a default low %ile (seems likely much higher than a 50th). If C-NDs are given a 34th %ile, I think that would affect %iles at the bottom of the range, but shouldn't affect the top %iles. If they get a 67th %ile, maybe not much of an effect.
Toy mice
🙋♂️Domestic short hair from a shelter. Fetched as a young kitten (a few months old), and is now almost 4 and still likes to play fetch. Smartest cat we’ve had, and is treat trained to do tricks.
Interesting. Did some sleuthing on Reporter, and 1) it looks like there are even some 1-year R01s, and 2) the majority of 2-year R01s are MYF (likely NIGMS trying to stretch funds while adhering to the letter of the law).
Based on total R35s funded in FY25 vs FY24, it looks like NIGMS tried to preserve MIRA success rates as much as possible by barely funding any R01s. But that’s not sustainable.
We are waiting for the break down by mech, and total application numbers, but this graph was based specifically on on R01s/R37s (but not R35s). Back of envelope says NIGMS funded about 7% of R01 applications. So 23.2% in 2023, 14.4% in 2024, to ~7% in 2025. Disaster.
MYF is hurting every IC, and the goalposts are shifting for MIRAs too. We just don’t have good data for how much.
Graph of award probability of R35 and R01 from NIH factbook as a function of review rank percentile. As is apparent, 2025 is a significant departure, with lower award probabilities at all scores <40 and significant departures from norm, where even being in the top 10% is no longer a nearly certain indicator of success. Data source: https://report.nih.gov/nihdatabook/report/302
The data is in: the NIH goalposts have shifted.
What were once almost certain fundable scores have become coin flips and what used to be likely grants have become aspirational, leading to fewer awards.
Another manifestation of how HHS policies have led to fewer awards and less science.
Here is the "effective payline" for each institute, estimated (by Claude) as the percentile where one can expect 80% probability of funding from a logistic regression fit. The effective payline has gone from a historic ~12% to 6% in 2025.
NIGMS looks to only be including R01s (a dwindling part of their portfolio), since R35s are not percentiled.
kallisto sped up RNA-seq quantification by 50x. Now another 50x speedup... quantify hundreds of millions of reads in a few seconds.
This seems too good to be true. But it's true!
Incredible accomplishment by @pmelsted.bsky.social.
🤯
Check out the new, improved version of our Mutation Browser preprint. We worked with a team from the Broad including their outreach office, a great summer intern, and the G2P project to add a new browser panel that maps mutations onto yeast protein structures! www.biorxiv.org/content/10.1...
We have started a project trying to predic the interactions/structures of all yeast protein pairs using an AlphaFold pooling approach. We are making the current dataset open and we welcome collaborations.
www.evocellnet.com/2026/03/mapp...
a headshot of David Botstein
We are deeply saddened to share the news of the passing of David Botstein, a towering figure in modern #genetics and a foundational force behind SGD.
www.yeastgenome.org/blog/in-memo... #yeast #modelOrganism
But things like claude-code really are an amplifier if you know what you are doing. And it is really freeing and wonderful to be able to focus more time on the most rewarding parts of this job. Having more time to think creatively, mentor our people...
This is the super hard part for comp bio. Students need a lot of up-front training in coding basics to be able write good prompts, interpret whether the resulting code is what they wanted, and especially, how to interpret the biology. Using AI to shortcut learning will lead to bad mistakes.
This part seemed very field-specific. Maybe for some fields where 'research assistants' are combing through books and papers researching a subject. Though in that case, AI is just a glorified search and summarization engine.
"we propose that the protoribosome was a parasite (...). If this view is correct, then like the spliceosome in the stem eukaryote, a repurposed host-parasite interaction led to a dramatic change in cell biology at the base of the tree of life, in this case leading to the exit from an RNA world"
Interesting, and this also just came out
academic.oup.com/pnasnexus/ar...
Provocative!
🤣
Leave is just treated as salary, so gets apportioned across sources of support. If it’s long enough to actually be a leave of absence, that’s a special category for effort reporting. For summer, we’re not allowed to take vacations during the exact times we’re getting paid off of grants.
Weird. We’re 9mo as well, but still get (and have to take) leave.
Made a short video as a response. 3 questions remain for me.
1) What is the definition of a "DEl activity," and how is it being communicated to NIH staff so they can determine whether a grant qualifies?
(Continued below)
I will go back and re-read Nancy Duarte's Resonate, and think about ways to make a better aspirational case. Because I'm in the same boat, as are many, and I appreciate Needhi starting these conversations.
I don't know the best way to balance these things. For those of us who've been at least moderately successful at this game, it's hard to know with very small n's whether we should be changing our strategies. At least more than just on the margins of what has already made us successful.
I've been chewing on this for days. I've gotten some comments where reviewers actually seemed to like my being nuanced and thoughtful. I *want* to be thought of as a careful scientist--that you can trust the shit that comes out of my lab. I also know that grants are aspirational--a call to action.
