OpenProtein.AI

?? Protein Revolution: The Tiny Model Making a Massive Impact! PoET-2 is changing the game in computational protein design, slashing experimental data needs by 30x! ?? learn more: https://lnkd.in/gVEHqeS8 #ProteinDesign #BiotechInnovation #AIRevolution

Announcing PoET-2: A breakthrough in protein engineering that delivers trillion-parameter performance with just 182M parameters We're excited to introduce PoET-2, our next-generation protein language model that transforms our ability to engineer proteins by integrating structural and evolutionary principles with a hyper-efficient multimodal architecture. Key breakthroughs: - Achieves state-of-the-art sequence and structure understanding with 500x less compute than contemporary models - Requires 30x less experimental data than existing methods for driving protein optimization - Enables accurate prediction of both substitutions and insertions/deletions, with or without experimental data - Unprecedented control over protein generation through a powerful new prompt grammar Real-world impact: - Optimize multiple protein properties simultaneously, with minimal experimental data - Design novel proteins using structure or function - inverse folding, motif scaffolding, and more - Reduce protein engineering cycles from months to weeks - Fast inference means faster results and the ability to explore more design space Read our technical deep-dive to learn how PoET-2 breaks conventional scaling laws and enables new possibilities in protein engineering: https://lnkd.in/gpTu3WuA PoET-2 is available today through the OpenProtein.AI platform: - Free for academic use - Full access to zero-shot predictions and property modeling - Python client and APIs for integration - Web interface for no-code access #ProteinEngineering #MachineLearning #Biotech #DrugDiscovery #Science

Our latest blog post explores PoET, our generative protein language model, as a foundation model for protein ML. With only 200 million parameters, PoET is significantly faster than models like ESM2 and xTrimoPGLM. Despite its size, PoET shines in fine-tuning experiments. PoET delivers 15x greater data efficiency when learning from small mutagenesis datasets and also allows us to improve signal peptide prediction in larger scale fine-tuning. PoET embeddings also capture significant structural information, from sequence alone, giving improved secondary structure prediction accuracy. Read more: https://lnkd.in/gUrM-zzi

Engineering protein complexes or multimeric proteins is challenging, especially with current approaches that only work with one subunit at a time. The OpenProtein.AI platform addresses these limitations, allowing you to model critical interactions between subunits easily. Dive into a step-by-step example of how to design libraries for multimeric proteins in our latest walkthrough. We demonstrate how to use the OpenProtein.AI platform to chain together subunits and design variants from a larger, more relevant design space. The walkthrough covers structure prediction and optimizing variant libraries. To highlight the platform’s flexibility to work with multimeric proteins, we designed a library of variants that incorporate data from heavy and light chains of an antibody, identifying high affinity binding variants with mutations across both chains. Read it here: https://lnkd.in/gMeHK5xq

Exciting News from OpenProtein.AI! We're thrilled to introduce the latest platform enhancements, including the AlphaFold2 Web Interface for visualizing predicted structures of multimers and single-chain sequences with precision. Our Batch Predictions feature now allows you to analyze multiple sequences using your trained models, streamlining your workflow. Plus, our Python Documentation has a fresh look, offering improved organization and walkthroughs for enzymes, antibodies, and structural proteins. Sign up for early access now: https://lnkd.in/gMmKKCfQ

Had an insightful conversation with Neeraj Rao discussing mutating proteins in-silico and learning about OpenProtein.AI .AI. Wish I had this tool during my lab work; it could have saved me lots of heartbreak. Looking forward to more insights at Future Labs Live Basel

"This study underscores the efficacy of evolutionary models in guiding large sequence alterations to generate functional diversity for protein design applications. Modifying enzymes or proteins while retaining functionality has been a major challenge. Often, multiple mutations or modifications to the primary structure of the enzyme have significant and destructive consequences to the 3D structure and functionality of the resulting enzyme. New methods to leverage the known information and predict the structure and function of designer proteins are necessary for progressing efficiently in protein modification studies." At OpenProtein.AI, we make it easy to use evolutionary models that explore exponentially more sequence space than traditional mutagenesis approaches so you can identify novel variants with a greater chance of success faster. Contact us today to learn how you can harness the power of evolution to create better proteins!

OpenProtein.AI will be exhibiting at Future Labs Live in two weeks!?This is Europe's biggest lab innovation event, happening in Basel on June 26-27th. Stop by startup booth S240 to see a demo of OpenProtein.AI in action and see how you can easily incorporate machine learning into your protein engineering workflow. #FutureLabsLive #machinelearning #proteinengineering #labautomation

OpenProtein.AI will be exhibiting at Future Labs Live!?This is Europe's biggest lab innovation event, happening at the Congress Center Basel on June 25-27th. We'll be demoing the OpenProtein.AI web app and PoET, our?generative protein language model for controllable sequence generation and zero-shot protein mutagenesis. Secure your free pass today and visit our booth in the startup zone. #FutureLabsLive #proteinengineering #machinelearning

PoET is our generative protein language model for controllable sequence generation, zero-shot protein mutagenesis & clinical variant effect prediction. In our latest blog post, we dive into PoET’s performance, which surpasses existing models in accuracy, speed, controllability, and generalizability. Bring the power of generative AI and ML to your protein engineering efforts today! No experimental data or coding experience needed. Find the link to our latest blog post in the comments below. Stay tuned for our upcoming post on PoET embeddings, also available on the OpenProtein.AI platform. #proteinengineering #machinelearning #openprotein #zeroshotengineering #MLforbiology #AIinMedicine #generativeAI