we've seen customers reaching back to physics based methods in order to solve their molecular modeling problems. In our latest post we explore how you can use the Rosetta "pocket opener" method with Boltz-2 to generate better protein-ligand models than Boltz-2 alone levitate.bio/boltz/ai/ml/...
How to train your model.
This immersive, hands-on, advanced-level event is designed for researchers who are ready to move beyond “out-of-the-box” machine learning models for protein design. Join us for this Summer RosettaCon 2025 workshop/hackathon, more here: rosettacommons.org/2025/05/30/h...
if you're interested in trying out our implementation of ipSAE along with the rest of our platform, get in touch and mention this post for a free 1 month trial levitate.bio/contact
As @rolanddunbrack.bsky.social recently discovered, one of the issues with the ipTM score used AF2 to represent the ipTM score is that it is biased by disordered regions of the protein outside the binding interface. To address this, he implemented a new metric, ipSAE.
levitate.bio/alphafold/mu...
there are still some real downsides to not actually modeling the physics
De novo #ai protein binder design with levitate.bio! Watch this video to see how it is done, with no code, no scripts, all in an easy to use GUI in the browser. youtu.be/87U_g8awaEE?...
BindCraft is a new AI de novo binder design tool from the Correia lab with 10-100% success rates. It's available today as an easy to use API using our API framework so you can deploy it in your own pipeline today.
Contact us at levitate.bio/contact to get started
www.biorxiv.org/content/10.1...
RFDiffusion is great for de novo design but it can be a bit frustrating to use. Check our our new GUI for making de novo binder design fast and easy! www.youtube.com/watch?v=rwpL...
Come work with us! We're hiring a Scientific Sales Specialist. This is the perfect job if you've got a BS in biology or biochemistry and are looking to start a career in software sales. No sales experience is necessary. For more details on the job and how to apply go to levitate.bio/careers
I sometimes forget how flexible proteins are. It becomes really apparent when you use NMR states for an animation. That poor cofactor is getting pushed around a lot 😅
#sciart #blender3d #biocatalysis
Quiz: Side chains R K E Q I M have 9 chi1+chi2 rotamers. They are:
{g+g+} {g+t} {g+g-} {tg+} {tt} {tg-} {g-g+} {g-t} {g-g-}.
4 of these have high energy & v. low populations compared to rest. Which 4? Why?
Leu has only 2 good rotamers. Which ones? why?
Good struct bio final exam Question.
Beyond excited to introduce AlphaQubit, now published in
Nature!
AlphaQubit is a neural network for quantum error correction and achieves state-of-the-art accuracy on simulated and real-world data.
Figure 3 from the preprint. its a 9 panel figure, and the figure caption makes for decent alt text so here it is: (A) AfCycDesign predicted model for design GAB_D8 bound to GABARAP shown as surface, with hotspot residues highlighted in green. (B) Affinity determination of GAB_D8 using SPR. SPR sensorgram from 9-point single-cycle kinetics experiments (5-fold dilution, highest concentration 20 µM). Experimental data are shown in orange and global fits are shown with black lines. The dissociation constant (KD) is also shown on the plot. (C) Superposition of chains E and F from the X-ray crystal structure of GAB_D8 bound to GABARAPL1 and the AfCycDesign model. (D) AfCycDesign predicted model for design GAB_D23 bound to GABARAP shown as surface, with hotspot residues highlighted in green. (E) Affinity determination of GAB_D23 using SPR. SPR sensorgram from 9-point single-cycle kinetics experiments (5-fold dilution, highest concentration 20 µM). Experimental data are shown in pink and global fits are shown with black lines. The dissociation constant (KD) is also shown on the plot. (F) Alignment of chains A and B from the X-ray crystal structure of GAB_D23 bound to GABARAP and the AfCycDesign model. (G) Alignments of GAB_D8 and GAB_D23 macrocycle models to X-ray crystal structures show close matches. (H) Comparison of GAB_D8 and GAB_D23 binding modes in the design models. (I) Competitive AlphaScreen response vs. concentration plot, IC50 from the average of three experiments. Donor and acceptor beads in the assay are bound to GABARAP and GABARAP-binding peptide K1, respectively.
This allows the models to "understand" the cyclic aspect of a cyclic peptide. And it seems to work pretty well! Decent looking designs within < 1Å of the crystal structure and <1nM IC50s. The code and models arent out yet but we look forward to tryin git out.
Here's an interesting new paper from IPD: www.biorxiv.org/content/10.1...
Cyclic peptides are a compelling format for a drug but a lot of computational design tools don't handle them well. Here they modified AF2 and RFDiffusion with a rotational encoding 1/
We're obviously going to use this account to talk about what we do, but we're also going to use it to share developments in other parts of the field that we think are particularly interesting
Hello! We're a software company which spun out of Cyrus Bio. We sell cloud based tools to help make state of the art protein engineering and modeling methods more accessible. We're owned by the Rosetta Commons Foundation which will use our profits to support academic research.
levitate.bio
