Align’s tournaments connect computational modeling directly to high-throughput experimentation, creating a shared arena for testing and improving protein engineering models. By challenging participants with real-world biological tasks, we generate benchmarks that help the community understand what works, what doesn’t, and where improvements are needed.
Tournaments allow us to
Measure the impact of new data and modeling paradigms on protein engineering performance
Create tight feedback loops between computation and experiments
Set shared goals for generative protein design across the research community
Each tournament is designed to tackle complex biological problems, from predicting the growth requirements of uncultured microbes to designing novel, functional enzymes. Competitions unfold in iterative rounds, where computational predictions and designs are tested against ground-truth experimental data. Tasks are selected not only for their scientific significance, like enabling plastic degradation, but also for their ability to support rapid co-development between models and lab experiments.
Competitions Catalyze Innovation
Historically, benchmarks are powerful tools for driving research breakthroughs.
Benchmarking tournaments have a long history of propelling progress in computational modeling and deep learning. Tournament platforms like ImageNet, Kaggle, DARPA Grand Challenges and CASP have been incredibly successful at connecting diverse machine learning researchers to high quality data sources. These tournaments have built communities around competition as they gained critical mass: they evolved from benchmarks for small but rigorously evaluated tasks to platforms for consistently catalyzing transformative scientific breakthroughs.
CASP stands out as a model for how carefully designed benchmarks, paired with high-quality experimental data, can transform methods in machine learning. For decades, CASP built a community of tournament participants and a platform for model evaluation and benchmarking. The Protein Data Bank (PDB) has a long history of building a large and diverse database of experimentally determined protein structures. Eventually the convergence of this rich and large experimental data source with CASPs increasingly sophisticated models led to the creation of AlphaFold2, which took an enormous bite out of the long-standing and extremely difficult protein structure prediction problem in both computation and biology.
How It Works
Each Tournament includes two rounds:
1. Predictive Phase
Participants predict functional properties of protein sequences. These predictions are scored against experimental data.
Top teams from the predictive round are invited to advance, but others may apply to participate directly in the generative phase.
2. Generative Phase
Teams design new protein sequences with desired traits. Designs are synthesized, tested in vitro, and ranked based on experimental performance.
Tournament Timeline
The Tournament runs approximately every 18–24 months, with new target proteins and expanded datasets in each edition. For example:
2023 Pilot: Enzyme function prediction and design across 6 donated datasets
2025 Tournament: Engineering improve PETase enzymes to tackle plastic waste
Ways to Participate
Whether you’re a model builder, a data generator, or an experiment designer, there’s a way to get involved:
Compete: Form a team and submit predictions or designs
Sponsor: Donate prizes, DNA synthesis, or cloud resources
Contribute a dataset: Host a new challenge track
Join the planning committee: Help shape the next Tournament
Contact tournament@alignbio.org to get started.