Prism Science

jobs.ashbyhq.comSenior Scientist, Structural Data RepresentationABOUT ASTERA: Astera is a private foundation on a mission to steer science and technology toward a better future. We believe the coming years will bring an era of unprecedented scientific and…

How should conformational ensembles be encoded for sharing, visualization, and reuse? Prism is hiring a Senior Scientist, Structural Data Representation to lead schemas, databases, and tool integration. Apply: jobs.ashbyhq.com/astera/2e65a...

jobs.ashbyhq.comSenior Scientist, Ensemble–Function MappingABOUT ASTERA: Astera is a private foundation on a mission to steer science and technology toward a better future. We believe the coming years will bring an era of unprecedented scientific and…

Which features of a conformational ensemble help predict protein function? Prism is hiring a Senior Scientist, Ensemble–Function Mapping to lead work on allostery, catalysis, and binding specificity. Apply: Senior Scientist, Ensemble–Function Mapping jobs.ashbyhq.com/astera/b5272...

sbgrid.orgSampleworks - SBGrid Consortium - Supported SoftwareThe SBGrid Consortium is an innovative global research computing group operated out of Harvard Medical School. SBGrid provides the global structural biology community with support for research…

Sampleworks is now on SBGrid. It combines biomolecular structure predictors with X-ray density maps to guide sampling of protein conformational ensembles. Explore Sampleworks: sbgrid.org/software/tit...

thestacks.orgIterating diffuse scattering protocols in parallel leads to reproducible, high-quality data from Mac1Two teams measured three-dimensional diffuse scattering maps from SARS-CoV-2 Nsp3 macrodomain (Mac1) crystals to hone protocols and assess reproducibility. The resulting high-quality maps and new…

Two teams, two facilities: Mac1 diffuse maps largely agree within measurement error. Study: thestacks.org/publications...

prismscience.orgTTPD: Invisible StringDynamic structural data, models, and language to become standard across the field. At scale, this is how we predict conformational ensembles across all of biology.

AI agents as research papers, and AI’s role in discovery: on the reading list for “Invisible String,” the new podcast from Stephanie Wankowicz and Jaime Fraser. Listen: prismscience.org/research/ttp...

prismscience.orgHow Silent Code Updates Can Alter Protein PredictionsDynamic structural data, models, and language to become standard across the field. At scale, this is how we predict conformational ensembles across all of biology.

Benchmarking protein predictors? Track preprocessing alongside weights. Feng Yu explains. Read here: prismscience.org/research/cod...

Prism Science reposted

We had an incredible few days down in Santa Cruz talking about the amazing science being done within @prismscienceorg.bsky.social!

huggingface.coPrismScienceOrg (Prism Science)We are building the foundation to make protein motion measurable, predictable, and actionable.

Protein motion needs a connected stack: measurement, processing, modeling, representations, and interpretation. We’ve set up a Prism space on Hugging Face as a home for future open-source work and scientific resources as they develop. Follow along here: huggingface.co/PrismScience...

prismscience.orgResearchResearch updates, scholarly work, code and datasets from the Prism team. Filter by topic and by the kind of output you are after.

Prism’s Research page now has an RSS feed. If RSS is how you keep up with papers, methods, and technical work, you can now add our work on protein motion and conformational ensembles to your reader. prismscience.org/research

prismscience.orgVarying crystallization conditions to study lattice effects on diffuse scatteringDynamic structural data, models, and language to become standard across the field. At scale, this is how we predict conformational ensembles across all of biology.

18 Sept

Microseeding made a big difference in Katie Lu’s Mac1 crystallization screens: 8 potential hits became 54. She’s using those results to optimize crystals for experiments asking how the crystal environment affects diffuse scattering. prismscience.org/research/kat...

Prism Science reposted

podcasts.apple.comPrism: Unlocking Protein DynamicsPodcast Episode · The Tortured Proteins Department · September 8 · 27m

9 Sept

Emergency 🚨 TTPD with @fraserlab.com about the launch of @prismscienceorg.bsky.social. podcasts.apple.com/us/podcast/p...

Prism Science reposted

prismscience.orgPrism | Revealing the Dynamics of LifeDynamic structural data, models, and language to become standard across the field. At scale, this is how we predict conformational ensembles across all of biology.

Proteins are always moving. They wiggle. They jiggle. They interact with their surroundings. That movement drives their function. Yet the motion stays mostly invisible. We cannot understand protein function, engineer it, or correct it until we can see the motion and predict it. prismscience.org

We have not yet tapped the full potential of cryoEM. The method revolutionized structural bio with atomic-level detail, but has largely focused on proteins’ common lowest-energy states. With enough samples, cryoEM sees a broader energy landscape: It can capture *rare* states.

View full thread

We envision that shifting data banks and experimental designs from static assumptions to dynamic expectations could change biology forever.

prismscience.orgOur ScienceStatic structures cannot tell us what makes a designed enzyme functional, what drives drug efficacy and resistance, or how to engineer motion itself.

Prism’s work will help reveal how a mutation shifts binding, stability, or activity, uncovering how function and disease actually work. Read more: prismscience.org/science/

But why care about rare? Rare and excited protein states are inherently difficult to detect. Yet they have outsized effects on actual protein function. Prism will use the strengths of cryoEM to discover these transient states and decode how proteins move between alternating shapes.

How will we use this data? Rare state data from cryoEM and other methods helps us build a new foundation for structural biology based on protein motion. Our team, led by Joey Davis (MIT), is creating new, public models for interpreting motion, and these methods will only improve with time.

We need new ways to understand disease & design drugs. Protein motion holds the answer.

Last century, structural biology relied on a motionless view of proteins. Today’s instruments & ML no longer have to. Prism will make protein visible, revealing how form creates function. See more: prismscience.org

Prism is building the full stack of technologies and algorithms necessary to make protein motion measurable, actionable, and predictable. Coordinated change to transform the field of structural biology.

diffuse.scienceCoffee is for Closers: making Mac1 diffuse scattering reproduciblediffuse scattering reproducible, yay!

4 Sept

Diffuse scattering is hard to compare quantitatively across experiments, let alone across synchrotrons. A new DiffUSE Project update shows diffuse maps reproducing across CHESS and ALS. Read the project update: diffuse.science/posts/reprod...

diffuse.scienceRevealing hidden signals in raw X-ray crystallography data with mdx2Mdx2 is a powerful and open-source toolkit for diffuse data processing.

3 Sept

Diffuse scattering can encode correlated protein motions, but standard pipelines often discard it as background. mdx2 is an open-source Python toolkit for carrying that signal through processing. Read @meisborg.bsky.social's overview and try mdx2: diffuse.science/posts/mdx2/

@biorxiv-biophys.bsky.social

1 Sept

Same mission, but an expanded vision for revolutionizing structural biology to better capture protein motion. Launching September 8.

Prism Science reposted

bioRxiv Biophysics

27 Aug

WaterFlow: Prediction of Ordered Water Molecule Positions on Protein Structures https://www.biorxiv.org/content/10.64898/2026.08.26.747373v1

biorxiv.orgWaterFlow: Prediction of Ordered Water Molecule Positions on Protein StructuresOrdered water molecules mediate many protein functions, including stability, ligand binding, and catalysis. Predicting their positions with sub-angstrom accuracy would support protein design, binding ...

28 Aug

Check out our latest! 🌊

Stephanie Wankowicz@stephanieaw.bsky.social

New pub from @diffuseproject.bsky.social! Ordered water molecules mediate protein stability, ligand binding, and catalysis. Predicting them lags far behind protein structure prediction. We developed WaterFlow, a new state-of-the-art method for ordered water placement. www.biorxiv.org/content/10.6...

21 Aug

DiffUSE is rebuilding GOODVIBES, a forward model for lattice vibrations in diffuse scattering, in fully open-source Python that handles any space group or unit cell. Stephen Thornton posts the progress and open questions. diffuse.science/posts/goodvi... #openscience

thestacks.orgRecovering Conformational Heterogeneity from the Protein Data Bank at ScaleAn Ensemble Dataset of over 60,000 Structures

11 Aug

Most structures in the PDB report one set of coordinates. The experimental data behind them, in both X-ray crystallography and cryo-EM, is produced by an ensemble. New work recovers that hidden signal at scale in over 60,000 structures. thestacks.org/publications...

11 Aug

We ran qFit across high-resolution PDB entries with deposited structure factors: 60,000+ ensemble models, the largest experimentally derived set to date. 83.7% fit the data better than the deposited structure, surfacing side-chain heterogeneity that single models miss.

thestacks.orgRecovering Conformational Heterogeneity from the Protein Data Bank at ScaleAn Ensemble Dataset of over 60,000 Structures

11 Aug

This reframes the public PDB as a source of experimental ensemble information, and gives ensemble prediction the training data it has been missing. Read the pub: thestacks.org/publications...

diffuse.scienceDisplaying Conformational Heterogeneity in BANDICOOTIn preparation for hierarchical hetereogeneity encoding, BANDICOOT has a new, more intuitive display mode for alternate conformations.

10 Aug

New in BANDICOOT: color each altloc a distinct hue, so you see heterogeneity while you build. Each added alternate conformation is auto hue-shifted from the last; remove them all and the residue returns to bulk color. Configurable and open source! diffuse.science/post/bandico... #compbio

diffuse.sciencemmCIF Explorer and the Heterogeneity ProposalA playground for the PDBx/mmCIF file exploration, a look at the current proposal for encoding correlated heterogeneity

29 July

An mmCIF file can say a residue has two alternate conformations and how often each appears. It cannot say which alternates occur together in the same copy. A draft proposal for recording that correlation, plus a tool to inspect it: diffuse.science/posts/mmcif-...

xtraj.pyIs it all just the lattice?Towards a complete forward model of diffuse scattering

27 July

In our latest blog, @fraserlab.com graduate student Taylor Womack discusses his roadmap before the results: how he plans to separate lattice vibrations from internal protein motion, which open tools he is using, and which parts are still unresolved. Feedback welcome. diffuse.science/posts/taylor...

22 July

New TTPD episode discusses two bioRxiv preprints, revisits the Bethesda Declaration one year on, and public pushback to the OMB funding rule. Plus: where blogs end and peer review begins. diffuse.science/posts/TTPD-16/ #StructuralBiology #Biophysics #OpenScience

diffuse.scienceBANDICOOT: Coot 0.9 That Runs on MacOS TahoeReviving Coot 0.9 for MacOS Tahoe as Bandicoot, a bespoke and integrated molecular modeling environment.

22 July

Bandicoot reengineers Coot 0.9 for macOS Tahoe: toolbars rebuilt, UI moved from Python 2 to C++, atom picking and labels reworked. Now with built-in pandda.inspect for examining partial-occupancy and alt conformations. diffuse.science/post/bandico...

diffuse.scienceIntroducing PLUG: A Compiler for Leakage-free Protein Function BenchmarkingLeakage-free Training to Predict Protein Function

15 July

Why do protein-function AI models look great on benchmarks but disappoint in practice? Often "leakage": proteins meant for testing slip into training data over time, inflating scores. PLUG is a tool to build cleaner training sets. Read Clay's work: diffuse.science/post/plug-be... #proteinML

github.comdiffUSEdiffUSE has 15 repositories available. Follow their code on GitHub.

1 July

Unlocking protein dynamics requires infrastructure we all can access. ICYMI: all of our software packages live on GitHub, updated in the open: github.com/diff-use

30 June

Open science is useful science. One effort we are developing are logbooks where we share research as it happens: what works, what doesn't, what we've learned. Led by DiffUSE Scientist Steve Meisburger (@meisborg.bsky.social). See the latest: diffuse.science/logbook/beam... #OpenScience

29 June

What's happening in protein dynamics research right now? Jaime & Stephanie break down NIH policy changes, preprints & the conformational ensembles conference. New episode up: diffuse.science/posts/TTPD-15/ #StructuralBiology

thestacks.orgthestacks.org

17 June

Bulk solvent fills ~half a protein crystal. Modeling this is required for a good fit to experimental data. How best to model this remains an open question. New from diffUSE: a Phenix patch to supply your own solvent map (MD, 3D-RISM, EDEN, neural nets). Read more here: thestacks.org/publications...

github.comGitHub - jmholton/phenix_user_solvent: allow user to override default solvent with externally-supplied mtz fileallow user to override default solvent with externally-supplied mtz file - jmholton/phenix_user_solvent

17 June

In a test case, swapping Phenix's flat bulk-solvent mask for an MD-derived map lowered cross-validated R-free from 0.147 to 0.109 in one 0.96 Å. Code here: github.com/jmholton/phe...

doi.orgsampleworks: A Modular Platform for Experimentally Guided Biomolecular Ensemble GenerationStructural biology increasingly relies on machine learning-based structure predictors, enabling accurate atomic-level structure prediction at scale. However, predicting conformational ensembles…

15 June

We share science at The DiffUSE Project in a few ways: blogs to contextualize motivation and intuition, scholarly outputs for methods, data, code, and a citable DOI. All open and shared rapidly for the best science. Our latest: doi.org/10.82153/jkx...

15 June

See our most recent blog diffuse.science/post/science... #SciComm #OpenScience

9 June

We present sampleworks, an open, modular platform for generating and evaluating biomolecular conformational ensembles from structure predictors guided by experimental data doi.org/10.82153/jkx....

github.comGitHub - diff-use/sampleworks: Framework for modified sampling from biomolecular generative modelsFramework for modified sampling from biomolecular generative models - diff-use/sampleworks

9 June

This open platform was built to allow further development of additional methods: github.com/diff-use/sampleworks Initial findings demonstrated that different structural predictors recover alternative conformations unevenly across proteins, suggesting memorization over generalization.

jobs.ashbyhq.comProject Manager - DiffUSE ProjectABOUT DIFFUSE The DiffUSE Project is building open infrastructure to study protein dynamics using experimental data. The project spans wet-lab collaborators, computational scientists, ML researchers,…

3 June

Want to shape the future of structural biology? We are seeking a Project Manager to help lead the DiffUSE Project, guiding ML researchers, computational scientists, national labs, and university partners. Apply: buff.ly/gZlj9AP #StructuralBiology #MachineLearning #Hiring #AcademicSky

Prism Science reposted

Gabriel Rocklin

@grocklin.bsky.social

18 May

Replying to @grocklin.bsky.social

Simulations can be valuable but there is a huge need for scalable experiments that probe protein dynamics- of course we're expanding our work here and there's also great work from others @hkws.bsky.social @ginaelnesr.bsky.social Basile Wicky (not on bsky) @diffuseproject.bsky.social

diffuse.scienceSubmit Examples of Heterogeneous Structural DataThe DiffUSE Project

28 May

The DiffUSE Project wants examples where structural heterogeneity matters but cannot be encoded in PDBx/mmCIF. X-ray, cryo-EM, time-resolved, multi-map, or fragment screening data welcome. Submit yours: diffuse.science/submit/

13 May

ICYMI: The latest Tortured Proteins Department episode. @stephanieaw.bsky.social and @fraserlab.com on: NIH strategic plan comments (due 5/26), AI lab guidelines, two new cryo-EM preprints, training scientists for what's next diffuse.science/posts/TTPD-14/

diffuse.scienceThe DiffUSE ProjectUnlocking Protein Dynamics

9 Apr

🎉🎉🎉

Stephanie Wankowicz@stephanieaw.bsky.social

I'm ecstatic to share that I'm joining the @diffuseproject.bsky.social (diffuse.science) full-time as Scientific Program Director at @asterainstitute.bsky.social .

Prism Science reposted