New paper showing that much of the apparent success of protein language models in predicting mutational effects is a mirage: These models mostly memorize sites. 1/
www.biorxiv.org/content/10.6...
It's a great morning for having my first senior author paper accepted 🥰🥰
April 1 deadline for @nhm.org Student Collections Study Award. Get $$ to spend time collecting data in our collections, including @tarpits.org
Open to:
- current undergraduate and grad students
- *not* local to Los Angeles (international applicants welcome)
More: nhm.org/student-coll...
Wow!!
Deadline for applying is 9 march. Don't miss out!
Cover of book ‘Ask for it’ Ask For It: How Women Can Use the Power of Negotiation to Get What They Really Want by Linda Babcock and Sara Laschever is a guide that teaches women negotiation skills for professional and personal situations, offering a four-phase program to help them identify their worth, strategize, and overcome common barriers like the fear of asking. The book uses real-life stories to illustrate how to ask for raises, promotions, and more, emphasizing that negotiation is essential and can be done in a way that feels comfortable for women.
My academic coach (a very cool privilege I have) just told me about ‘Ask For It’ a guide for women to negotiate a raise. It came up after I mentioned equity in pay 💰 in academia.
The number of times I've been a reviewer and had to ask folks to report their sample sizes
I've attended a conference (with 2 exceptions) every year since 2011.
COOL!
Soon we will have vegan AI, no animals were harmed in the training of this model
love being a WOC pointing out zero representation and getting the academic version of "all lives matter" back. "we don't see color" isn't the flex you think it is. it's why nothing changes.
One of my Math colleagues pointed out that there was still an increase in students enrolled in remedial math just prior to the calculator ban (from 2020 to 2023). Anything beyond that is hard to decipher.
For me, outside of policy, this makes clear how impactful 2020 was for students :/
They were amazing talks!!! Very well deserved.
Congratulations!!
I've definitely considered it. But i agree that there's a massive sunk cost (considering we are responding to suggested revisions).
This is the front cover of the new edition of @thelancet.com (thanks to @profstevegriffin.bsky.social for sharing) #HealthPolicy #Science 🧪🧵
Hmm… so this means that colleges that want to “prepare students for the AI workplace” should be sure to keep robust Romance and other language departments?
Worth a shot.
#AcademicSky
NSF leaders have just acknowledged what many scientists have long suspected: Presidential directives to boost AI and quantum have upended its traditional way of doing business. www.science.org/content/arti...
Median amount of time spent under review is 7.4–14.6% longer for female-authored articles than for male-authored articles and the differences remain significant after controlling for several factors - analysis of >36.5 million articles in >36,000 journals
doi.org/10.1371/jour...
My current record is a manuscript that has been in review for like a year.
And that only includes a single round of reviews so far
I had a pre-interview meeting to discuss a (then) open postdoc position. I asked the PI if they prioritize publishing with the high school students they work with and their response was:
"We don't give token authorship"
This was at an R1.
One thing I've been thinking about is whether we should use the 0.5 cutoff as a standard.
It's our informal verbal standard, so why not prompt people to perform hypothesis tests where the ICC is tested against 0.5 instead of 0?
3 samples per animal is amazing!
0.5 is acceptable, fitting the common assumption of greater differences between individuals (or species) than within.
Our single sample estimates had ICC = 0.21. But averaging just 2 (intraindividual) estimates got our average repeatability up to 0.5 🦎 (Fig 1C)
The take home message:
Any measure of intra-individual variation is better than none. Please sample individuals at least N=2 times per treatment if you want to avoid this error variance from being funneled into your parameter estimates and statistical tests!
In this paper we collected about 100 individual sprints per animal! (they are so cute look at them!).
We used Intraclass Correlation Coefficients and bootstrapping to discuss alternative ways to analyze maximum performance data, with specific attention on the reliability of the resulting studies 😎
🚨 NEW preprint!!! 🚨
Browning, Mayor, and the rest of the Fall 2025 Vert Anat class at OWU are pleased to present you with our study:
Maximum performance, repeatability, and intraindividual variability of sprinting in common wall lizards (Podarcis muralis) doi.org/10.32942/X29...
The take home message:
Any measure of intraindividual variation is better than none. Please sample individuals at least N=2 times per treatment if you want to avoid this error variance from being funneled into your parameter estimates and statistical tests!
In this paper we collected about 100 individual sprints per animal! (they are so cute look at them!).
We used Intraclass Correlation Coefficients and bootstrapping to discuss alternative ways to analyze maximum performance data, with specific attention on the reliability of the resulting studies 😎
A tortoise beetle on a leaf. The clear elytra extend past the body of the critter. The body is blue and yellow and iridescent.
Sometimes there are jewels just sitting on leaves out here
This is a tortoise beetle, beautifully iridescent, about the size of my fingernail.
