Alchemab

Generative antibody language model for antigen-binding prediction and de novo paired heavy-light chain design, scaled to 2.4 billion parameters.

Structure-aware antibody language model with sequence embeddings contrastively aligned to crystal structures, reaching AUROC 0.875 on HER2 binding.

Antibody language model pretrained on 779M human B cell receptor sequences, reading a heavy chain, a light chain, or both paired in one input.

BERT-based antibody language model pretrained on 57M B cell receptor sequences for paratope prediction and convergent antibody discovery.