TLDR: Chai Discovery Team presents Chai-2. A multimodal AI-model that allows zero-shot, de novo design of antibodies. Achieving a 16% hit rate across 52 novel targets using ≤20 candidates per target, Chai-2 outperforms prior methods by over 100x and delivers validated binders in under two weeks—eliminating the need for large-scale screening.
Chai’s Discovery Team introduced a new tool for drug discovery that is a breakthrough in computational methods. Chai-2The platform is capable of designing zero-shot antibodies and proteins. Chai-2 designs binders that are functional in less than a minute, as opposed to previous methods which relied on high throughput screening. Single 24-well Plate Setting up, achieving More than 100 fold improvement over existing state-of-the-art (SOTA) methods.
Chai-2 has been tested 52 novel targetsThis system was able to achieve a high level of performance despite the challenge. This was not a problem for the system. 16% experimental hit rateDiscovering binders within the first a Two-week cycle From computational design to validation in the wet lab. This performance represents a change from probabilistic screen to deterministic creation in molecular engineering.
AI-Powered De Novo Design at Experimental Scale
Chai-2 is an integrated system that enables you to integrate Chai-2 into your existing systems. all-atom generative design module The system operates in a cellular environment. The system is based on a zero-shot settingThe sequences can be generated for antibody types like scFvs (sequences of Fvs) and VHHs (vectors with high affinity), without the requirement of prior binders.
Chai-2 features include:
- No Target-Specific Tuning It is not required
- Ability prompt designs using epitope-level constraints
- Generating Therapeutically appropriate formats (miniproteins, scFvs, VHHs)
- Support for Cross-reactivity Design The difference between two species
This approach allows researchers to design ≤20 antibodies or nanobodies per target and bypass the need for high-throughput screening altogether.
Benchmarking Across Diverse Protein Targets
Chai-2 is a powerful tool for achieving high-quality lab results. No sequence or structural similarity with known antibodies. The designs were created and then tested with Bio-layer Interferometry Results show: Results are shown:
- 15.5% average hit rate All Formats
- VHHs get a 20.0% discount, 13 % for scFvs
- Success Binders for Twenty-six out of fifty targets were met
Chai-2 has been able to produce hits on hard targets, such as TNFαIn silico technology has been a challenge for many years. Many binders were shown Dissociation constants from low-nanomolar KDs to picomolarThe arrow indicates high affinity interactions.
Uniqueness, diversity, and specificity
The outputs of Chai-2 are both structurally and chronologically different from antibodies. The structural analysis revealed:
- There is no generated design.
- All CDR sequences had >10 edit distance from the closest known antibody
- Binders were found to be grouped in multiple clusters for each target. Conformational Diversity
Further evaluations have been confirmed Low off-target Binding You can also find out more about the following: Comparable polyreactivity profiles Trastuzumab (and Ixekizumab) are clinical antibodies.
Flexible Design and Customization
Chai-2 goes beyond general-purpose binding generation.
- Multiple Targets The epitopes found on one protein
- Binders across Different antibody formats (e.g., scFv, VHH)
- Generate Cross-species Reactive Antibodies In one simple prompt
An antibody designed by Chai-2 achieved cross-reactivity during a study. nanomolar KDs The utility of the antibody against human and cyno versions of a particular protein was demonstrated. Preclinical research and Therapeutic Development.
Drug Discovery: Implications
Chai-2 reduces the time required for traditional biotechnology discovery. Weeks to monthsThe system delivers leads that have been experimentally verified in just one go. The combination of a high success rate with a novel design and modular prompting represents a new paradigm in the therapeutic discovery workflows.
This framework is not limited to antibodies. miniproteins, macrocycles, enzymesPotentially, and. Small moleculesThe way to success is paved with a number of steps. computational-first design paradigms. Future directions include expanding into bispecifics, ADCsExplore the world with. Biophysical property optimization (e.g., viscosity, aggregation).
Chai-2, a generative model that uses AI to design molecules in real world drug discovery environments, sets a high bar as the AI field matures.
Click here to find out more Technical Report. The researchers are the sole owners of all credit. Also, feel free to follow us on Twitter, Youtube You can also find out more about the following: Spotify Join our Facebook group! 100k+ ML SubReddit Subscribe now our Newsletter.
Asif Razzaq, CEO of Marktechpost Media Inc. is a visionary engineer and entrepreneur who is dedicated to harnessing Artificial Intelligence’s potential for the social good. Marktechpost was his most recent venture. This platform, which focuses on machine learning and deep-learning news, is well known for being both technically and easily understood by the general public. Over 2 million views per month are a testament to the platform’s popularity.
