Top Tweets for #MoML2026
zPocket paper just won Best Paper at #MoML2026. Huge congratulations to our intern @talthegal and kudos to her mentors Allison Keys, Ben DiFrancesco and @gruver_nate from @genesismolai
Finding where a drug can bind on a protein is step zero of structure-based design. Most pocket finders look at the shape of a structure and guess which grooves look druggable. That works well enough on familiar proteins and much worse on new ones.
Turns out that co-folding models already have a good chunk of knowledge of where small molecules bind, because they have learned from a huge number of protein–ligand complexes.
In this work, a tiny head is trained on top of the frozen co-folding model weights. That head beat every pocket finder we compared against, on every slice of the test set.
And then of course the benchmarks: existing pocket benchmarks predate the training data of modern co-folding models. The team built PocketWatch, which only tests pockets that showed up after the training cutoff and sorts them by how new they really are. Truly novel pockets are harder for every method. zPocket still leads there.
Paper, code, weights and data are all open:
Paper: https://t.co/UwjAfhxDHC
Code: https://t.co/ZdvMVLL4Pt
Weights: https://t.co/YAMyzEWppf
Data: https://t.co/ZglpBUo7e2
Congratulations to Tally Portnoi, @MIT PhD student and one of our 2026 summer interns, on winning the Octavian-Eugen Ganea Prize for Best Paper for “zPocket: Finding Pockets with Frozen Co-Folding Features,” at #MoML2026 hosted by @AIHealthMIT.
Our interns take on real problems that matter to our research and to the wider scientific community. We love seeing that work earn the recognition it deserves.
Congratulations to Tally Portnoi, @MIT PhD student and one of our 2026 summer interns, on winning the Octavian-Eugen Ganea Prize for Best Paper for “zPocket: Finding Pockets with Frozen Co-Folding Features,” at #MoML2026 hosted by @AIHealthMIT.
Our interns take on real problems that matter to our research and to the wider scientific community. We love seeing that work earn the recognition it deserves.
Industry talk #2! Danaher’s Andrey Tovchigrechko talks about how his team used @boltz_bio’s BoltzGen to design beta-zipper mini binder interfaces while using agentic tools to optimize the in silico pass rate in BoltzGen. #MoML2026
How are industry experts using ML in their research? @ByteDanceSeed_’s @chengyenhsieh presents on Protein Autoregressive Modeling: Taking initiative from the hierarchical nature of proteins, PAR generates backbones by forming a global topology & performing refinements. #MoML2026
Up next, @YaleMed’s @MarkGerstein talks about building world models to predict protein folding pathways 🧬 #MoML2026
Starting now, @ISTAustria’s Alex Bronstein talks about developing a general inference time optimization framework that can maximize log-likelihood and incorporate external constraints while eliminating dependence on diffusion length and removing initialization bias. #MoML2026
Now hearing from @FU_Berlin’s @CecClementi on the short-sightedness of GNNs and potential solutions. She introduces a framework for efficiently and accurately simulating biomolecular complexes. #MoML2026
#MoML2026 Top 4: @PaulSharoubeem presents Electro-Prot, 6.7M-parameter encoder-decoder transformer trained on 162,980 APBS calculated fields to predict Poisson-Boltzmann potentials at arbitrary points around a protein!
#MoML2026 Top 4: @Northeastern and @broadinstitute’s Yikun Zhang presents on MolSAGE, a framework that accumulates and and reuses experience across tasks to build state-of-the-art drug models efficiently!
👀 Next up: @pas_saro and @NoahBGetz from @boltz_bio present on their approaches to maximizing performance on structure and affinity prediction in the Boltz model series, and their work on using BoltzMol-1 to identify a target inhibiting DNA damage repair. #MoML2026
#MoML2026 is underway! #JameelClinic faculty lead @BarzilayRegina opens the conference with a presentation on the state-of-the-art in molecular modeling.
We're excited to hear from @YaleMed's @MarkGerstein on Wednesday, Oct. 14 for #MoML2026! <20 tickets left!
🎟️ Register now before tickets are sold out: https://t.co/lzl4hTohHY
We're excited to welcome @FU_Berlin's @CecClementi to MIT for #MoML2026 on Oct. 14!
🎟️ <30 tickets left! Register now: https://t.co/0396INwdoA
We're excited to announce that @pas_saro, co-founder of @boltz_bio, will be delivering the keynote for #MoML2026! 🎟️ Only 10% of seats remain — register now to secure your spot: https://t.co/lzl4hTohHY
Happy to share that our work on "Dissecting the limitations of AlphaFold3 confidence metrics for de novo antibody design selection", was accepted as a poster at #MoML2026.
I'll be at MIT on Oct 14 -- looking forward to meeting others working on protein design!
@AIHealthMIT
⏱️Good news for anyone racing against the clock: the #MoML2026 deadline has been extended to Sept. 11 — 1 more week to submit! If you're looking for sharp feedback, new collaborators, & a chance to present your work at a leading institution, this is your window of opportunity ⬇️
⏰ REMINDER: Today is the LAST DAY for paper submissions to #MoML2026! First authors w/ accepted papers get FREE admission and four will have the chance to present their work in the Top 4! One of these four will take away the Best Paper Prize!
Submit now: https://t.co/IZS9yyrd1P
<1 month left before the paper submission deadline for #MoML2026 on Sept. 1! First authors w/ accepted papers will receive FREE admission! 🏆 4 papers will be selected for oral presentations, but only one will win the Best Paper Prize! More info in 🧵⬇️
Last Seen Hashtags on Sotwe
mgai(*)****************(*)(*)(*)********* filter:native_video
volleyballsetup
Seen from Japan
男同
Seen from France
momson()+filter:videos
Seen from Netherlands
minichat
MGAI()Filter:Videos
Seen from Belgium
قحـبة
Seen from Saudi Arabia
minichat() #snapchat
Seen from Poland
scarystrength
Seen from United States
mgai
Seen from Turkey
Most Popular Users
Elon Musk 
@elonmusk
241.5M followers
Barack Obama
@barackobama
118.8M followers
Cristiano Ronaldo
@cristiano
114.5M followers
Donald J. Trump
@realdonaldtrump
111.8M followers
Narendra Modi
@narendramodi
107.1M followers
Rihanna
@rihanna
98.7M followers
NASA
@nasa
92.3M followers
Justin Bieber
@justinbieber
91.8M followers
KATY PERRY
@katyperry
90M followers
Taylor Swift
@taylorswift13
84M followers
Lady Gaga
@ladygaga
75.5M followers
Virat Kohli
@imvkohli
73.5M followers
Kim Kardashian
@kimkardashian
70.9M followers
YouTube
@youtube
68.8M followers
Neymar Jr
@neymarjr
66.5M followers
Bill Gates
@billgates
65.3M followers
Selena Gomez
@selenagomez
63.1M followers
The Ellen Show
@theellenshow
62.2M followers
CNN
@cnn
61.7M followers
X
@x
60.7M followers