Machine Learning in Structural Biology

Invited Speakers

Erika Alden DeBenedictis

Group Leader, Francis Crick Institute

Show/Hide Bio

Gabe Rocklin

Assistant Professor, Department of Pharmacology, Northwestern University.

Show/Hide Bio

Jennifer Listgarten

Professor in EECS, UC Berkeley

Show/Hide Bio

Milot Mirdita

Postdoctoral Researcher, Seoul National University.

Show/Hide Bio

Noelia Ferruz

Group Leader, Center of Genomic Regulation, Barcelona.

Show/Hide Bio

Josh Abramson, AlphaFold3

Senior ML Researcher, Google DeepMind.

Show/Hide Bio

Jeremy Wohlwend, Boltz-1

PhD Student, Massachusetts Institute of Technology.

Show/Hide Bio

Schedule (Pacific Standard Time)

08:40
08:30	Opening Remarks
08:35	Invited Speaker - Noelia Ferruz Tailored protein design with conditional language models
08:45
08:50
08:55
09:00	Contributed Talk The OMG dataset: An Open MetaGenomic corpus for mixed-modality genomic language modeling Yunha Hwang · Andre Cornman · Jacob West-Roberts · Martin Beracochea · Sergey Ovchinnikov · Simon Roux · Antonio Camargo · Milot Mirdita
09:05
09:10
09:15	Invited Speaker - Josh Abramson, AlphaFold3 Team Biomolecular Structure Prediction with AlphaFold3
09:20
09:25
09:30
09:35
09:40	Break
09:45
09:50
09:55	Invited Speaker - Erika Alden DeBenedictis How do we make the next PDB? Introducing Align’s Data Platform
10:00
10:05
10:10
10:15
10:20	Contributed Talk Controllable All-Atom Generation of Protein Sequence and Structure from Sequence-Only Inputs Amy Lu · Wilson Yan · Kevin Yang · Vladimir Gligorijevic · Kyunghyun Cho · Richard Bonneau · Pieter Abbeel · Nathan Frey
10:25
10:30
10:35	Contributed Talk Protein Language Model Fitness is a Matter of Preference Cade Gordon · Amy Lu · Pieter Abbeel
10:40
10:45
10:50	Invited Speaker - Gabe Rocklin Learning the determinants of protein folding energy landscapes from large-scale experiments
10:55
11:00
11:05
11:10
11:15	Word from Sponsors
11:20	Poster Session/Lunch
11:25
11:30
11:35
11:40
11:45
11:50
11:55
12:00
12:05
12:10
12:15
12:20	Invited Speaker - Jennifer Listgarten Protein-Protein Interactions in Synthetic Co-Evolution & Other Stories
12:25
12:30
12:35
12:40
12:45	Contributed Talk Generative modeling of protein ensembles guided by crystallographic electron densities Sai Advaith Maddipatla · Nadav Sellam · Sanketh Vedula · Ailie Marx · Alex Bronstein
12:50
12:55

01:05
01:00	Contributed Talk Improving Ab-Initio Cryo-EM Reconstruction with Semi-Amortized Pose Inference Shayan Shekarforoush · David Lindell · Marcus Brubaker · David Fleet
01:10
01:15	Break
01:20
01:25
01:30	Invited Speaker - Milot Mirdita Challenges to metagenomic annotation with Foldseek and protein language models
01:35
01:40
01:45
01:50
01:55	Invited Speaker - Jeremy Wohlwend, Boltz-1 Boltz-1: Democratizing Biomolecular Interaction Modeling
02:00
02:05
02:10
02:15
02:20	Contributed Talk FlowPacker: protein side-chain packing with torsional flow matching Jin Sub Lee · Philip Kim
02:25
02:30
02:35	Contributed Talk HelixFlow, SE(3)–equivariant Full-atom Design of Peptides With Flow-matching Models Xuezhi Xie · Pedro A Valiente · Jisun Kim · Jin Sub Lee · Philip Kim
02:40
02:45
02:50	Break
02:55	Break
03:00	Word from PLINDER / PINDER competition
03:05	PLINDER / PINDER competition talks
03:10	PLINDER / PINDER competition talks
03:15	Panel Session
03:20
03:25
03:30
03:35
03:40
03:45
03:50
03:55	Closing Remarks
04:00	Poster Session / Happy Hour
04:05
04:10
04:15
04:20
04:25
04:30
04:35
04:40
04:45
04:50
04:55
05:00

Accepted Papers

LatentDE: Latent-based Directed Evolution accelerated by Gradient Ascent for Protein Sequence Design

Thanh Tran, Nhat Khang Ngo, Duy Nguyen, Truong Son Hy

[paper]
Assessing interaction recovery of predicted protein-ligand poses

Frederic Dreyer, David Errington, Cedric Bouysset, Constantin Schneider

[paper]
Improving Inverse Folding models at Protein Stability Prediction without additional Training or Data

Oliver Dutton, Sandro Bottaro, Michele Invernizzi, Istvan Redl, Albert Chung, Falk Hoffmann, Louie Henderson, Stefano Ruschetta, Fabio Airoldi, Benjamin M J Owens, Patrik Foerch, Carlo Fisicaro, Kamil Tamiola

[paper]
Improving Antibody Design with Force-Guided Sampling in Diffusion Models

Paulina Kulyte, Francisco Vargas, Simon Mathis, Yuguang Wang, Jose Miguel Hernandez-Lobato, Pietro Liò

[paper]
Equivariant Blurring Diffusion for Multiscale Generation of Molecular Conformer

Jiwoong Park, Yang Shen

[paper]
Active Learning for Affinity Prediction of Antibodies

Alexandra Gessner, Sebastian Ober, Owen Vickery, Dino Oglic, Talip Ucar

[paper]
IgBlend: Unifying 3D Structure and Sequence for Antibody LLMs

Cedric Malherbe, Talip Ucar

[paper]
Learning the Language of Protein Structures

Jeremie DONA, Benoit Gaujac, Timothy Atkinson, Liviu Copoiu, Thomas Pierrot, Thomas Barrett

[paper]
moPPIt: De Novo Generation of Motif-Specific Binders with Protein Language Models

Tong Chen, Yinuo Zhang, Zachary Quinn, Pranam Chatterjee

[paper]
Improving generalisability of 3D binding affinity models in low data regimes

Julia Buhmann, Ward Haddadin, Alan Bilsland, Lukáš Pravda, Hagen Triendl

[paper][preprint]
Active Learning for Energy-Based Antibody Optimization and Enhanced Screening

Kairi Furui, Masahito Ohue

[paper]
Conditional Enzyme Generation Using Protein Language Models with Adapters

Jason Yang, Aadyot Bhatnagar, Jeffrey Ruffolo, Ali Madani

[paper][preprint]
Improving Structural Plausibility in 3D Molecule Generation via Property-Conditioned Training with Distorted Molecules

Lucy Vost, Vijil Chenthamarakshan, Payel Das, Charlotte Deane

[paper]
Understanding Protein-DNA Interactions by Paying Attention to Protein and Genomics Foundation Models

Dhruva Abhijit Rajwade, Erica Wang, Aryan Satpathy, Alexander Brace, Hongyu Guo, Arvind Ramanathan, Shengchao Liu, Animashree Anandkumar

[paper]
SPECTRE: A Spectral Transformer for Molecule Identification

Wangdong Xu

[paper]
FlowPacker: protein side-chain packing with torsional flow matching

Jin Sub Lee, Philip Kim

[paper]
Similarity-Quantized Relative Difference Learning for Improved Molecular Activity Prediction

Karina Zadorozhny, Kangway Chuang, Bharath Sathappan, Ewan Wallace, Vishnu Sresht, Colin Grambow

[paper]
HERMES: Holographic Equivariant neuRal network model for Mutational Effect and Stability prediction

Gian Marco Visani, Michael Pun, William Galvin, Eric Daniel, Kevin Borisiak, Utheri Wagura, Armita Nourmohammad

[paper]
Adapting protein language models for structure-conditioned design

Jeffrey Ruffolo, Aadyot Bhatnagar, Joel Beazer, Stephen Nayfach, Jordan Russ, Emily Hill, Riffat Hussain, Joseph Gallagher, Ali Madani

[paper][preprint]
Allo-Allo: Data-efficient prediction of allosteric sites

Tianze Dong, Christopher Kan, Kapil Devkota, Rohit Singh

[paper][preprint]
CryoSPIN: Improving Ab-Initio Cryo-EM Reconstruction with Semi-Amortized Pose Inference

Shayan Shekarforoush, David Lindell, Marcus Brubaker, David Fleet

[paper]
GFlowNet Pretraining with Inexpensive Rewards

Mohit Pandey, Gopeshh Subbaraj, Artem Cherkasov, Emmanuel Bengio

[paper]
Benchmarking text-integrated protein language model embeddings and embedding fusion on diverse downstream tasks

Young Su Ko

[paper][preprint]
RNAgrail: graph neural network and diffusion model for RNA 3D structure prediction

Marek Justyna

[paper]
The OMG dataset: An Open MetaGenomic corpus for mixed-modality genomic language modeling

Yunha Hwang, Andre Cornman, Jacob West-Roberts, Martin Beracochea, Sergey Ovchinnikov, Simon Roux, Antonio Camargo, Milot Mirdita

[paper]
Functional Alignment of Protein Language Models via Reinforcement Learning with Experimental Feedback

Nathaniel Blalock, Srinath Seshadri, Philip Romero, Agrim Babbar, Sarah Fahlberg

[paper]
Antibody Library Design by Seeding Linear Programming with Inverse Folding and Protein Language Models

Conor Hayes, Andre Goncalves, Steven Magana-Zook, Ahmet Solak, Daniel Faissol, Mikel Landajuela

[paper]
EpiGraph: Recommender-Style Graph Neural Networks for Highly Accurate Prediction of Conformational B-Cell Epitopes

Jung-Eun Shin, Yen-Lin Chen, Nathan Rollins, Thomas Hopf, Jordan Anderson, Michael Cianci, Daniela Cipolletta, Jyothsna Visweswaraiah, Yi Xing, Colin Lipper, Kevin Otipoby, Nathan Higginson-Scott, Ryan Peckner

[paper]
MF-LAL: Drug Compound Generation Using Multi-Fidelity Latent Space Active Learning

Peter Eckmann, Dongxia Wu, Germano Heinzelmann, Michael Gilson, Rose Yu

[paper]
Higher-Order Message Passing for Glycan Representation Learning

Roman Joeres, Daniel Bojar

[paper]
LOCAS: Multi-label mRNA Localization with Supervised Contrastive Learning

Abrar Abir, Md Toki Tahmid, M. Saifur Rahman

[paper]
Does Structural Information Improve ESM3 for Protein Binding Affinity Prediction?

Thomas Loux, Dianzhuo Wang

[paper]
Unified Sampling and Ranking for Protein Docking with DFMDock

Lee-Shin Chu, Sudeep Sarma, Jeffrey Gray

[paper][preprint]
Expanding Automated Multiconformer Ligand Modeling to Macrocycles and Fragments

Jessica Flowers

[paper][preprint]
Protein Sequence Domain Annotation using Language Models

Arpan Sarkar, Kumaresh Krishnan, Sean Eddy

[paper]
ProtComposer: Compositional Protein Structure Generation with 3D Ellipsoids

Hannes Stärk, Bowen Jing, Tomas Geffner, Jason Yim, Tommi Jaakkola, Arash Vahdat, Karsten Kreis

[paper]
Rapid protein structure assessment via a forward model for NMR spectra

Benjamin Harding, Chad Rienstra, Hannah Wayment-Steele, Ziling Hu, Frank Delaglio, Rajat Garg, Katherine Henzler-Wildman, Timothy Grant

[paper]
PropEn: Optimizing Proteins with Implicit Guidance

Natasa Tagasovska, Vladimir Gligorijevic, Kyunghyun Cho, Andreas Loukas

[paper][preprint]
Bayesian Optimisation for Protein Sequence Design: Gaussian Processes with Zero-Shot Protein Language Model Prior Mean

Carolin Benjamins, Shikha Surana, Oliver Bent, Marius Lindauer, Paul Duckworth

[paper]
Bio-xLSTM: Generative modeling, representation and in-context learning of biological and chemical sequences

Niklas Schmidinger, Lisa Schneckenreiter, Philipp Seidl, Johannes Schimunek, Sohvi Luukkonen, Pieter-Jan Hoedt, Johannes Brandstetter, Andreas Mayr, Sepp Hochreiter, Guenter Klambauer

[paper][preprint]
Retrieval Augmented Protein Language Models for Protein Structure Prediction

Peter Lee, Xavier Cheng, Eric Xingyi, Le Song

[paper]
CompassDock: A Comprehensive Accurate Assessment Approach for Deep Learning-Based Molecular Docking in Inference and Fine-Tuning

Ahmet Sarigun, Vedran Franke, Bora Uyar, Altuna Akalin

[paper][preprint]
BoostMD: Accelerating Molecular Sampling using ML Force Field Feature

Lars Schaaf, Ilyes Batatia, Christoph Brunken, Thomas Barrett, Jules Tilly

[paper]
DockFormer: Efficient Multi-Modal Receptor-Ligand Interaction Prediction using Pair Transformer

Ben Shor, Dina Schneidman

[paper]
Cryo-EM images are intrinsically low dimensional

Luke Evans, Octavian-Vlad Murad, Lars Dingeldein, Pilar Cossio, Roberto Covino, Marina Meila

[paper]
Open-source Tools for CryoET Particle Picking Machine Learning Competitions

Kyle Harrington, Zhuowen Zhao, Jonathan Schwartz, Saugat Kandel, Utz Ermel, Mohammadreza Paraan, Clinton Potter, Bridget Carragher

[paper]
Protein Language Model Fitness is a Matter of Preference

Cade Gordon, Amy Lu, Pieter Abbeel

[paper]
Balancing Locality and Reconstruction in Protein Structure Tokenizer

Jiayou Zhang, Barthélémy Meynard, Jing Gong, Xavier Cheng, Eric Xing, Le Song

[paper]
What has AlphaFold3 learned about antibody and nanobody docking, and what remains unsolved?

Fatima Hitawala, Jeffrey Gray

[paper]
HelixFlow, SE(3)–equivariant Full-atom Design of Peptides With Flow-matching Models

Xuezhi Xie, Pedro A Valiente, Jisun Kim, Jin Sub Lee, Philip Kim

[paper]
MolMix: A Simple Yet Effective Baseline for Multimodal Molecular Representation Learning

Andrei Manolache, Dragos-Constantin Tantaru, Mathias Niepert

[paper]
Integrating Macromolecular X-ray Diffraction Data with Variational Inference

luis aldama, Kevin Dalton, Doeke Hekstra

[paper]
Fine-Tuning Discrete Diffusion Models via Reward Optimization: Applications to DNA and Protein Design

Chenyu Wang, Masatoshi Uehara, Yichun He, Amy Wang, Tommaso Biancalani, Avantika Lal, Tommi Jaakkola, Sergey Levine, Hanchen Wang, Aviv Regev

[paper][preprint]
Low-N OpenFold fine-tuning improves peptide design without additional structures

Theo Sternlieb, Jakub Otwinowski, Sam Sinai, Jeffrey Chan

[paper]
SPRINT Enables Interpretable and Ultra-Fast Virtual Screening against Thousands of Proteomes

Andrew McNutt, Abhinav Adduri, Caleb Ellington, Monica Dayao, Eric Xing, Hosein Mohimani, David Koes

[paper]
Ranking protein-peptide binding affinities with protein language models

Charles Chalas, Michael Dunne, Michael Dunne, charles chalas

[paper]
Generating and scoring stable proteins using joint structure and sequence modeling

Yehlin Cho, Justas Dauparas, Kotaro Tsuboyama, Gabriel Rocklin, Sergey Ovchinnikov

[paper]
FusOn-pLM: A Fusion Oncoprotein-Specific Language Model via Adjusted Rate Masking

Sophia Vincoff, Shrey Goel, Kseniia Kholina, Rishab Pulugurta, Pranay Vure, Pranam Chatterjee

[paper]
Systems-Structure-Based Drug Design

Vincent Zaballa, Elliot Hui

[paper]
Learning the language of protein-protein interactions with ESM-Multimer

Varun Ullanat, Bowen Jing, Samuel Sledzieski, Dr. Bonnie Berger

[paper]
Guided Multi-objective Generative AI for Structure-based Drug Design

Amit Kadan, Kevin Ryczko, Erika Lloyd, Adrian Roitberg, Takeshi Yamazaki

[paper]
Tradeoffs of alignment-based and protein language models for predicting viral mutation effects

Noor Youssef, Sarah Gurev, Navami Jain, Debora Marks

[paper]
IgFlow: Flow Matching for De Novo Antibody Design

Sanjay Nagaraj, Amir Shanehsazzadeh, Hyun Park, Jonathan King, Simon Levine

[paper]
Generating and evaluating diverse sequences for protein backbones

Yo Akiyama, Sergey Ovchinnikov

[paper]
SuperMetal: A Generative AI Framework for Rapid and Precise Metal Ion Location Prediction in Proteins

Xiaobo Lin, Zhaoqian Su, Yunchao Liu, Jingxian Liu, Xiaohan Kuang, Peter Cummings, Jesse Spencer-Smith, Jens Meiler

[paper]
Antibody DomainBed: Out-of-Distribution Generalization in Therapeutic Protein Design

Natasa Tagasovska, Ji Won Park, Matthieu Kirchmeyer, Nathan Frey, Andrew Watkins, Aya Ismail, Arian Jamasb, Edith Lee, Tyler Bryson, Stephen Ra, Kyunghyun Cho

[paper][preprint]
Controllable All-Atom Generation of Protein Sequence and Structure from Sequence-Only Inputs

Amy Lu, Wilson Yan, Kevin Yang, Vladimir Gligorijevic, Kyunghyun Cho, Richard Bonneau, Pieter Abbeel, Nathan Frey

[paper]
Loop-Diffusion: an equivariant diffusion model for designing and scoring protein loops

Kevin Borisiak, Gian Marco Visani, Armita Nourmohammad

[paper]
ProteinZen: combining latent and SE(3) flow matching for all-atom protein generation

Alex Li, Tanja Kortemme

[paper]
Exploring Discrete Flow Matching for 3D De Novo Molecule Generation

Ian Dunn, David Koes

[paper]
SynFlowNet: Design of Diverse and Novel Molecules with Synthesis Constraints

Miruna Cretu, Charles Harris, Ilia Igashov, Arne Schneuing, Marwin Segler, Bruno Correia, Julien Roy, Emmanuel Bengio, Pietro Lió

[paper]
RNA-DCGen: Dual Constrained RNA Sequence Generation with LLM-Attack

Haz Sameen Shahgir, Md. Rownok Zahan Ratul, Md Toki Tahmid, Khondker Salman Sayeed, Atif Rahman

[paper][preprint]
RNA-GPT: Multimodal Generative System for RNA Sequence Understanding

YIJIA XIAO, Edward Sun, Yiqiao Jin, Wei Wang

[paper]
TomoPicker: Annotation-Efficient Particle Picking in Cellular cryo-electron Tomograms

Mostofa Rafid Uddin, Ajmain Yasar Ahmed, Toki Tahmid, Alam, Min Xu

[paper]
AptaBLE: An Enhanced Deep Learning Platform for Aptamer Protein Interaction Prediction and Design

Sawan Patel, Sherwood Yao, Zhangzhi Peng, Keith Fraser, Pranam Chatterjee, Adam Friedman, Owen Yao

[paper]
Capturing Protein Dynamics: Encoding Temporal and Spatial Dynamics from Molecular Dynamics Simulations

Vignesh Bhethanabotla, Amin Tavakoli, Animashree Anandkumar, William Goddard

[paper]
ProPicker: Promptable Segmentation for Particle Picking in Cryogenic Electron Tomography

Simon Wiedemann, Zalan Fabian, Mahdi Soltanolkotabi, Reinhard Heckel

[paper]
Estimating protein flexibility via uncertainty quantification of structure prediction models

Charlotte Sweeney, Nele Quast, Fabian Spoendlin, Yee Whye Teh

[paper]
Generative modeling of protein ensembles guided by crystallographic electron densities

Sai Advaith Maddipatla, Nadav Bojan Sellam, Sanketh Vedula, Ailie Marx, Alex Bronstein

[paper]
Energy-Based Flow Matching for Molecular Docking

Wenyin Zhou, Christopher Sprague, Hossein Azizpour

[paper]
Controlling multi-state conformational equilibria of dynamic proteins with Frame2seq

Deniz Akpinaroglu, Dominic Grisingher, Stephanie Crilly, Tanja Kortemme

[paper]

Challenge 2024

This year we are running a challenge on the Pinder and Plinder datasets to evaluate how well the community is currently doing for protein-protein interaction prediction and protein-ligand complex prediction.

Details of the challenge

To submit your trained model you will need to make an inference docker image on HuggingFace Spaces using the following templates:

Rules for model training

Participants MUST use the sequences and SMILES in the provided train and validation sets from PINDER or PLINDER. In order to ensure no leakage, external data augmentation is not allowed.
If starting structures/conformations need to be generated for the model, then this can only be done from the training and validation sequences and SMILES. Note that this is only the case for train & validation - no external folding methods or starting structures are allowed for the test set under any circumstance!. Only the predicted structures/conformers themselves may be used in this way, the embeddings or models used to generate such predictions may not. E.g. it is not valid to “distill” a method that was not trained on PLINDER/PINDER
The PINDER and PLINDER datasets should be used independently; combining the sets is considered augmentation and is not allowed.
For inference, only the inputs provided in the evaluation sets may be used: canonical sequences, structures and MSAs; no alternate templates or sequences are permitted. The inputs that will be used by assessors for each challenge track is as follows:
- PLINDER: (SMILES, monomer protein structure, monomer FASTA, monomer MSA)
- PINDER: (monomer protein structure 1, monomer protein structure 2, FASTA 1, FASTA 2, MSA 1, MSA 2)
Model selection must be performed exclusively on the validation set designed for this purpose within the PINDER and PLINDER datasets.
Methods relying on any model derivatives or embeddings trained on structures outside the PINDER/PLINDER training set are not permitted (e.g., ESM2, MSA: allowed; ESM3/ESMFold/SAProt/UniMol: not allowed).

Please find the technical documentation for how to use the datasets for the challenge:

Rules for valid inference pipeline

Submission system will use Hugging Face Spaces. To qualify for submission, each team must:

Provide an MLSB submission ID or a link to a preprint/paper describing their methodology. This publication does not have to specifically report training or evaluation on the P(L)INDER dataset. Previously published methods, such as DiffDock, only need to link their existing paper. Note that entry into this competition does not equate to an MLSB workshop paper submission.
Create a copy of the provided inference template. Go to the top right corner of the page and click on the drop-down menu (vertical ellipsis) right next to the “Community”, then select “Duplicate this space”.
Change files in the newly create space to reflect the peculiarities of your model
- Edit requirements.txt to capture all dependencies.
- Include a inference_app.py file. This contains a predict function that should be modified to reflect the specifics of inference using their model.
- Include a train.py file to ensure that training and model selection use only the PINDER/PLINDER datasets and to clearly show any additional hyperparameters used.
- Provide a LICENSE file that allows for reuse, derivative works, and distribution of the provided software and weights (e.g., MIT or Apache2 license).
- Modify the Dockerfile as appropriate (including selecting the right base image)
Submit to the leaderboard via the designated form.
- On submission page, add reference to the newly created space in the format username/space (e.g mlsb/alphafold3)

How to submit and view results

Metrics

PLINDER: lDDT-PLI
PINDER: DockQ

Other metrics computed by PINDER/PLINDER will be displayed on the leaderboard but will not influence the ranking.

The winners will be invited to present their work at the MLSB workshop.

Evaluation Datasets

Although the exact composition of the eval set will be shared at a future date, below we provide an overview of the dataset and what to expect

Two leaderboards, one for each of PINDER and PLINDER, will be created using a single evaluation set for each.
Evaluation sets will be subsets of 150-200 structures from the current PINDER and PLINDER test splits (subsets to enable reasonable eval runtime).
Each evaluation sample will contain a predefined input/output to ensure performance assessment is model-dependent, not input-dependent.
The focus will be exclusively on flexible docking/co-folding, with a single canonical structure per protein, sampled from apo and predicted structures.
Monomer input structures will be sampled from paired structures available in PINDER/PLINDER, balanced between apo and predicted structures and stratified by "flexibility" level according to specified conformational difference thresholds.