Stephanie Wankowicz
@stephanieaw.bsky.social
Computational Structural Biologist
Assistant Professor @VanderbiltMPB
wankowiczlab.com
(she/her)
Past: UCSF, Dana-Farber, Broad Institute, UMass Amherst
Assistant Professor @VanderbiltMPB
wankowiczlab.com
(she/her)
Past: UCSF, Dana-Farber, Broad Institute, UMass Amherst
Actually the OPPOSITE.
November 21, 2025 at 8:14 PM
Actually the OPPOSITE.
Was thinking the same thing!
November 19, 2025 at 10:00 PM
Was thinking the same thing!
Preprint: Experiment-guided AlphaFold3 resolves accurate protein ensembles www.biorxiv.org/content/10.1...
Experiment-guided AlphaFold3 resolves accurate protein ensembles
AlphaFold3 predicts highly accurate protein structures from sequence, but tends to collapse to a single dominant conformation, even when the underlying structure is inherently heterogeneous. Moreover,...
www.biorxiv.org
October 23, 2025 at 3:39 PM
Preprint: Experiment-guided AlphaFold3 resolves accurate protein ensembles www.biorxiv.org/content/10.1...
Episode 8: Open Science in a Changing Academic World
open.spotify.com
October 23, 2025 at 3:39 PM
Feedback welcomed!
October 23, 2025 at 2:24 PM
Feedback welcomed!
3) Experimental approaches for determining ensembles face major challenges, including ensemble averaging, data sparsity, and intrinsic measurement errors.
4) Standardized representations, comparison metrics, and uncertainty quantification are still lacking.
Feedback welcomed!
4) Standardized representations, comparison metrics, and uncertainty quantification are still lacking.
Feedback welcomed!
October 23, 2025 at 2:24 PM
3) Experimental approaches for determining ensembles face major challenges, including ensemble averaging, data sparsity, and intrinsic measurement errors.
4) Standardized representations, comparison metrics, and uncertainty quantification are still lacking.
Feedback welcomed!
4) Standardized representations, comparison metrics, and uncertainty quantification are still lacking.
Feedback welcomed!
Yes, that points out another challenge about these models is how they are generated and what data they are generated should lead to different interpretations.
October 22, 2025 at 3:07 PM
Yes, that points out another challenge about these models is how they are generated and what data they are generated should lead to different interpretations.
Ensemble models can imply correlated motions, but they aren’t directly encoded. You need to derive covariance, but we could directly put this into the model (where supported by data).
October 22, 2025 at 12:54 PM
Ensemble models can imply correlated motions, but they aren’t directly encoded. You need to derive covariance, but we could directly put this into the model (where supported by data).
Fantastic question! Yes, the PDB format (multiconf and ensemble) is limited in showing correlated motion. We are working on integrating a new hierarchical encoding strategy that will help with this and is hopefully set to be approved by the mmCIF working group soon! journals.iucr.org/m/issues/202...
Comprehensive encoding of conformational and compositional protein structural ensembles through the mmCIF data structure
Traditional structural models of biomolecules typically represent only a single conformational state, even though biomolecules naturally exist in multiple states crucial for their function. Here, we p...
journals.iucr.org
October 21, 2025 at 8:03 PM
Fantastic question! Yes, the PDB format (multiconf and ensemble) is limited in showing correlated motion. We are working on integrating a new hierarchical encoding strategy that will help with this and is hopefully set to be approved by the mmCIF working group soon! journals.iucr.org/m/issues/202...
There are also some other great blog posts and logbooks on our initial data collections. Check out diffuse.science to see what we have been up to! @diffuseproject.bsky.social @fraserlab.com @andolab.bsky.social @nozomi-ando.bsky.social
The Diffuse Project
The Diffuse Project
diffuse.science
October 21, 2025 at 4:30 PM
There are also some other great blog posts and logbooks on our initial data collections. Check out diffuse.science to see what we have been up to! @diffuseproject.bsky.social @fraserlab.com @andolab.bsky.social @nozomi-ando.bsky.social
There is still a lot of development needed to make these useful, but in an effort to iteratively share our science, we wanted to release this as an initial start. Read more here: diffuse.science/posts/multi_... and feel free to test it out and/or provide feedback!
diffuse.science
October 21, 2025 at 4:30 PM
There is still a lot of development needed to make these useful, but in an effort to iteratively share our science, we wanted to release this as an initial start. Read more here: diffuse.science/posts/multi_... and feel free to test it out and/or provide feedback!