Unlock Antibody Design with LICHEN, Now Integrated into Vecura
Researchers can now leverage the LICHEN generative model directly within Vecura to design, generate, and score compatible antibody light-chain sequences without the need for complex local infrastructure.
What is LICHEN?
LICHEN (Light-chain Immunoglobulin sequence generation Conditioned on the Heavy chain and Experimental Needs) is a generative AI model designed to solve a core challenge in antibody discovery: identifying light-chain sequences that are immunologically compatible with a specific heavy chain. By learning the conditional distribution of light chains based on natural antibody pairings, the model can generate compatible partners or evaluate existing heavy/light chain combinations. It is particularly valuable for therapeutic antibody design, including the specialized requirements of bispecific antibody engineering.
What can users do with LICHEN on Vecura?
With LICHEN on Vecura, users can:
- Generate novel, compatible light-chain sequences for a given heavy chain, including options for bispecific designs by supplying two heavy chains.
- Steer generation processes by specifying preferred light-chain types, V-gene families, or grafting specific CDR sequences.
- Quantitatively evaluate and rank candidate heavy/light chain pairings using conditional log-likelihood and perplexity scores.
- Apply advanced filtering strategies—such as redundancy removal, diversity maximization, and humanness scoring—directly within the workflow.
What the output means
The output provides generated amino acid sequences for candidate light chains, alongside quantitative metrics including log-likelihood and perplexity. Log-likelihood measures the probability of the pairing under the model's learned distribution, while perplexity provides a per-token measure of the sequence's naturalness given the heavy chain.
These outputs should be used to support scientific decision-making, such as prioritizing candidates for laboratory screening. They do not replace experimental validation.
Why this matters
Finding the right light chain for a therapeutic heavy chain is a complex optimization problem that requires balancing binding affinity, stability, and "humanness" to minimize potential immunogenicity. LICHEN streamlines this process by providing a data-driven approach to partner discovery, reducing the experimental burden of testing non-functional or poorly folded pairings.
By enabling targeted generation and rigorous quantitative scoring, LICHEN allows researchers to navigate the vast antibody sequence space more efficiently, helping to accelerate the development of high-quality therapeutic candidates.
- Developed by: The Oxford Protein Informatics Group (OPIG)
- Source: Official GitHub Repository
- Reference: Communications Biology 2026
Try Lichen on Vecura.
Open the model workspace and start evaluating it with your own inputs.