Protein Engineering Tournament: A New Frontier
Scientists compete to design and test innovative proteins through collaboration.
Chase Armer, Hassan Kane, Dana L. Cortade, Henning Redestig, David A. Estell, Adil Yusuf, Nathan Rollins, Hansen Spinner, Debora Marks, TJ Brunette, Peter J. Kelly, Erika DeBenedictis
― 7 min read
Table of Contents
- The Challenge of Predicting Protein Behavior
- Generative Modeling: Crafting New Proteins
- A New Competition: The Protein Engineering Tournament
- Tournament Structure: How It Works
- The In Silico Round
- The In Vitro Round
- The Teams Take Part
- Datasets and Events: The Ingredients
- Results of the Tournament: The Bake-Off Outcomes
- The In Vitro Round Results
- Learning and Adapting: What Comes Next
- Conclusion: The Future of Protein Engineering
- Original Source
- Reference Links
The world of proteins can be pretty complex, but scientists are trying to make some sense of it all. They’re working hard on figuring out how proteins work and how to design new ones. Think of proteins as tiny machines in our bodies that help things happen-like digesting food or fighting off germs. However, figuring out how to make these machines work better is a bit like trying to bake a cake without a recipe.
The Challenge of Predicting Protein Behavior
Scientists use something called modeling to predict how proteins will behave based on their genetic code. It’s a bit like guessing what kind of cake you’ll get based on the ingredients you throw into a bowl. Unfortunately, for many proteins, the data needed to make these predictions is lacking. The available Datasets, like those in ProtaBank, are often too simple. They only look at basic changes in proteins, which makes it hard for scientists to accurately guess how these proteins will work in real life.
Imagine trying to guess how a cake will taste if all you know is that it has flour and sugar but no idea about the eggs or the baking time. That’s the current situation with protein datasets. Without diverse and complicated data, understanding proteins becomes a tricky business.
Generative Modeling: Crafting New Proteins
On the other hand, there’s another approach called generative modeling. This is where scientists aim to design new proteins with specific traits. They want to create proteins that might perform better than the ones we currently have, like proposing a new cake recipe that could taste even better.
However, there’s a snag. Scientists often struggle to test their ideas in real life. They lack the means to verify if the new protein designs actually work as planned. Think of it as whipping up a cake recipe but having no oven to bake it in. Without proper testing, it’s hard to know which recipes (or protein designs) are worth pursuing.
A New Competition: The Protein Engineering Tournament
To tackle these issues, a fun competition called the Protein Engineering Tournament was born. The plan is simple: bring scientists together to predict and create amazing proteins. By offering new datasets and opportunities for hands-on experience, the tournament hopes to bridge the gap between theory and practice.
Think of this tournament as a bake-off for scientists where they can show off their best protein recipes while competing with their peers. We’re talking about a friendly rivalry where researchers can test their skills and share knowledge. What’s not to love?
Tournament Structure: How It Works
The tournament consists of two main rounds: an in silico round and an in vitro round.
The In Silico Round
In the first round, teams can use computer models to predict how various protein sequences will behave. They are given a set of protein sequences and asked to guess traits like stability and activity. It’s similar to guessing how a cake will turn out based on the ingredients alone.
For this round, there are two tracks: zero-shot and supervised. In the zero-shot track, participants have to make predictions without any prior data. It’s a bit like trying to bake a cake for the first time without any instructions. Teams in this track had just a few weeks to come up with their best guesses.
The supervised track gives teams some training data to work with, which is a bit like giving them a basic recipe to follow. They can then train their models and see how well their predictions hold up against unseen data.
The In Vitro Round
The second round is where the fun really begins. Here, participants use their computational models to create new proteins that exhibit desirable traits. This is where the action happens-teams propose exciting new protein designs based on their predictions and test them out in labs.
The goal is to create proteins with improved activity while keeping stability in check. It’s not all about making the fanciest cake; it’s about ensuring it will still be delicious after it comes out of the oven.
The Teams Take Part
When the tournament kicked off, over 90 participants formed 28 teams, including folks from universities, companies, and independent researchers. Everyone was excited to see whose approach would lead to the tastiest proteins-uh, we mean, the best-performing protein models.
In the in silico round, teams submitted their predictions, and those who performed well were invited to move on to the in vitro round. Those teams got to play with the real deal: bringing their designs to life and seeing if they worked as well as they hoped.
Datasets and Events: The Ingredients
The tournament was made possible thanks to the generosity of academic and industry partners, who provided various datasets for the teams to work with. In total, there were six unique datasets, including one that was used multiple times.
These datasets were like the secret ingredients in a cake recipe, offering teams the chance to try different approaches. With these resources, participants could experiment and find the best solutions to the challenges posed during the competition.
Results of the Tournament: The Bake-Off Outcomes
After much anticipation, the results from the in silico round came in. Teams were ranked based on how accurately they predicted protein properties. The Marks Lab snagged the top spot in the zero-shot track, while Exazyme and Nimbus shared the glory in the supervised track. The overall champion for this round was Nimbus, who proved their modeling skills!
The In Vitro Round Results
The in vitro round was where things got really exciting. Teams had to present their new protein designs, and then those designs were tested in the lab. Imagine the suspense: watching your cake being put into the oven, waiting to see how it would rise.
During this round, TUM Rostlab stole the show with their top-scoring enzyme variant, while MediumBio and Marks Lab also showcased impressive designs. Most teams managed to create proteins that performed better than the originals, with some even exceeding expectations.
Learning and Adapting: What Comes Next
The tournament was a huge success, demonstrating that collaboration and competition can lead to meaningful advancements in protein engineering. It highlighted the importance of sharing knowledge, pooling resources, and fostering creativity.
Looking ahead, future tournaments will build upon this foundation, tackling more complex protein functions and incorporating lessons learned from the pilot event. The ultimate goal is to keep pushing the boundaries of what’s possible in protein design.
The Protein Engineering Tournament is more than just a competition-it’s a community-driven platform designed to accelerate the field of protein engineering. By creating opportunities for scientists to test their ideas and share their findings, this initiative paves the way for the next generation of protein research.
Want to see what came out of this scientific bake-off? All the data, results, and team submissions are available for anyone interested in jumping into the wonderful world of protein engineering. It’s like inviting everyone to the cake party, where everyone can try their hand at baking up something new!
Conclusion: The Future of Protein Engineering
With excitement brewing in the protein engineering community, the Protein Engineering Tournament has set the stage for future innovations. By fostering collaboration and competition, scientists can continue to learn from each other and push the boundaries of what’s possible with proteins.
Who knows what deliciously creative discoveries lie ahead? As scientists whip up new ideas and challenge each other, we can look forward to a future full of exciting breakthroughs in protein design and function. Get ready to dig into the next round of scientific fun!
Title: Results of the Protein Engineering Tournament: An Open Science Benchmark for Protein Modeling and Design
Abstract: The grand challenge of protein engineering is the development of computational models to characterize and generate protein sequences for arbitrary functions. Progress is limited by lack of 1) benchmarking opportunities, 2) large protein function datasets, and 3) access to experimental protein characterization. We introduce the Protein Engineering Tournament--a fully-remote competition designed to foster the development and evaluation of computational approaches in protein engineering. The tournament consists of an in silico round, predicting biophysical properties from protein sequences, followed by an in vitro round where novel protein sequences are designed, expressed and characterized using automated methods. Upon completion, all datasets, experimental protocols, and methods are made publicly available. We detail the structure and outcomes of a pilot Tournament involving seven protein design teams, powered by six multi-objective datasets, with experimental characterization by our partner, International Flavors and Fragrances. Forthcoming Protein Engineering Tournaments aim to mobilize the scientific community towards transparent evaluation of progress in the field. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=113 SRC="FIGDIR/small/606135v2_ufig1.gif" ALT="Figure 1"> View larger version (30K): [email protected]@11da734org.highwire.dtl.DTLVardef@1cc51c0org.highwire.dtl.DTLVardef@10b3c27_HPS_FORMAT_FIGEXP M_FIG C_FIG
Authors: Chase Armer, Hassan Kane, Dana L. Cortade, Henning Redestig, David A. Estell, Adil Yusuf, Nathan Rollins, Hansen Spinner, Debora Marks, TJ Brunette, Peter J. Kelly, Erika DeBenedictis
Last Update: 2024-11-19 00:00:00
Language: English
Source URL: https://www.biorxiv.org/content/10.1101/2024.08.12.606135
Source PDF: https://www.biorxiv.org/content/10.1101/2024.08.12.606135.full.pdf
Licence: https://creativecommons.org/licenses/by/4.0/
Changes: This summary was created with assistance from AI and may have inaccuracies. For accurate information, please refer to the original source documents linked here.
Thank you to biorxiv for use of its open access interoperability.