🧬Jacob L Steenwyk
banner
jlsteenwyk.bsky.social
🧬Jacob L Steenwyk
@jlsteenwyk.bsky.social
UC-Berkeley Postdoc🐻, Edison Scientific Consultant👨🏽‍💻, Evomics Workshop Codirector🧬

prev Vanderbilt PhD, FutureHouse, Latch, MantleBio (acquired)

🌲 https://linktr.ee/jlsteenwyk
📍 https://jlsteenwyk.com/
November 22, 2025 at 6:34 AM
Prompt: "Can you rotate the clade encompassing A, B, C and D"

1st image is the original, the 2nd image is after 'rotation'

They are the same. The LLM was unable to rotate the clade.

Frontier Labs, we can get this right - I feel very strongly about our ability to do so!
bsky.app/profile/jlst...
November 20, 2025 at 5:27 PM
Companies building frontier LLMs, please let me help you teach your models how to interpret evolution

These are *so* wrong, but I know we can get it right.

prompt: "create an image of a phylogenetic tree with the following topology (((A,B),(C,D)),E);"
November 20, 2025 at 5:42 AM
Over a century ago, Hannah Höch gifted us with the Dada piece titled "Cut with the Kitchen Knife Dada Through the Last Weimar Beer-Belly Cultural Epoch of Germany" (1919-20).

imo, this collage shreds meaning, mocks certainty, and still feels more honest than most timelines today. Awesome piece
November 17, 2025 at 3:43 AM
Cherry chip #mushrooms 🍒🍄

This reddish-orange #mushroom gets its name from a #cherry-like coloration🔴 and for being found in #wood chips🌳

Awesome find in the backyard!
November 16, 2025 at 1:20 AM
Tomorrow registration closes for the #Phylogenomics and Evolutionary #Genomics workshop in the beautiful Czech Republic!

The #faculty are world-class -- this isn't one to miss!

Register today and learn both the fundamentals and advanced techniques in the field!

evomics.org/apply-worksh...
November 14, 2025 at 7:06 PM
We cite Schultz et al several times in the manuscript (see Table 1 & Fig. 1, where it is coded as R23, and several other places in the manuscript).

We also cite relevant, recent literature that suggests synteny might not be the best phylogenetic marker - e.g., www.biorxiv.org/content/10.1...
November 14, 2025 at 12:11 AM
Thank you - thank you for multiple reasons!

Those images of the sponges are beautiful, and your insights on the manuscript were very helpful! Thank you for being so generous!
November 13, 2025 at 10:49 PM
Oh no! Do I have to read the novel to talk to them? Or....

(only joking around :D)
November 13, 2025 at 10:06 PM
In fig S22, we also conducted traditional topology tests using either concatenation (A) or coalescence (B) using the consistent genes and found that they also supported the sponge-sister hypothesis.

I thought this analysis was really cool and a great suggestion by a reviewer.
November 13, 2025 at 8:34 PM
(E & F) After reanalyzing 10 previously published datasets, we found all 170 significant tests supported the sponge-sister hypothesis, and there was no support for the ctenophore-sister hypothesis
November 13, 2025 at 8:34 PM
(C & D) I didn't mention it before, but we also made sure our consistent genes had robust phylogenetic signal. Reviewers asked us to explore what happens when that threshold is relaxed (thresholds [or Th] 2 through 5)

Relaxed thresholds led to more inconclusive test results
November 13, 2025 at 8:34 PM
(A & B) Each dataset (row) was also subsampled based on various metrics associated with robust phylogenomic signal & implemented amino acid recoding schemes (20, 18, 12 or 6 characters; columns)

135 tests supported the sponge-sister hypothesis; none supported the ctenophore-sister hypothesis
November 13, 2025 at 8:34 PM
A&B) New datasets in gray. Published datasets that supported the ctenophore- (orange) & sponge-sister hypothesis (blue) tended to have fewer taxa (x-axis) and shorter alignment lengths (y-axis).

C&D) Newly constructed datasets remained larger than previous ones after removing inconsistent genes
November 13, 2025 at 8:34 PM
We unified the 2 major phylogenomic methods (concatenation and coalescence) into a single workflow by measuring single-gene signal, identifying those w/ consistent signal between the two and discarding those that disagreed
November 13, 2025 at 8:34 PM
NEW pub in @science.org 🥳

Is it sponges (panels A & B) or comb jellies (C & D) that root the animal tree of life?

For over 15 years, #phylogenomic studies have been divided.

We provide new evidence suggesting that...

🔗: www.science.org/doi/10.1126/...
November 13, 2025 at 8:34 PM
It's okay, I'm good
November 13, 2025 at 2:58 AM
Here is the full list of instructors in case it would be helpful :)
November 12, 2025 at 4:02 AM
Feedback from the last #Phylogenomics & EvoGenomics workshop

1) The whole experience was fantastic!! Thank you for organising such an amazing workshop!

2) I’ve learned more here than I did in 4 years of my undergraduate studies

Register today! www.evomics.org/apply-workshop-on-phylogenomics-2026/
November 11, 2025 at 5:38 PM
forgot to attached the image lol
bsky.app/profile/jlst...
November 11, 2025 at 5:30 PM
The connections & friends I met at the Workshop on #Phylogenomics and #EvolutionaryGenomics are of immense personal and professional value

I keep in contact with a lot of folks from the workshop, and a lot of benefits came from beyond the classroom

Apply by Nov 15
evomics.org/apply-worksh...
November 11, 2025 at 4:27 PM
At the workshop on #Phylogenomics & #EvolutionaryGenomics, @ankarn.bsky.social will present work on #microbial #ecology & #phylogenetics

We are incredibly excited that a former workshop director will be returning to share her latest work!

Register by Friday, Nov 15!
evomics.org/apply-worksh...
November 11, 2025 at 4:22 PM
Phylogenomics and Evolutionary Genomics Workshop in Czechia - 5 days until registration closes!

evomics.org/apply-worksh...
November 10, 2025 at 8:10 PM
The full list of instructors is available here
bsky.app/profile/jlst...
November 10, 2025 at 7:28 PM
At the workshop on #Phylogenomics and #EvolutionaryGenomics, Frederik Ronquist will present a lecture & lab on probabilistic programming in phylogenetics

Ensure you can expand your skills at this workshop and register by Nov 15!
evomics.org/apply-worksh...
November 10, 2025 at 6:31 PM