Co-Diffusion: An Affinity-Aware Two-Stage Latent Diffusion Framework for Generalizable Drug-Target Affinity Prediction

A research team led by Yining Qian has introduced Co-Diffusion, an innovative AI framework that redefines drug-target affinity (DTA) prediction as a constrained latent denoising process. The model addresses a critical limitation in current deep learning approaches: representation collapse in 'cold-start' regimes where limited labeled data and domain shifts prevent learning transferable pharmacophores and binding motifs. Co-Diffusion employs a two-stage paradigm where Stage I establishes an affinity-steered latent manifold by aligning drug and target embeddings under explicit supervision, ensuring the latent space reflects the intrinsic binding landscape.

Stage II introduces modality-specific latent diffusion as a stochastic perturb-and-denoise regularizer, forcing the model to recover consistent affinity semantics from noisy structural representations. This approach effectively mitigates the reconstruction-regression conflict common in generative DTA models. Theoretically, the researchers show that Co-Diffusion maximizes a variational lower bound on the joint likelihood of drug structures, protein sequences, and binding strength.

Extensive experiments across multiple benchmarks demonstrate that Co-Diffusion significantly outperforms state-of-the-art baselines, particularly yielding superior zero-shot generalization on unseen molecular scaffolds and novel protein families. This breakthrough paves a robust path for in silico drug prioritization in unexplored chemical spaces, potentially reducing the time and cost of early-stage drug discovery by enabling more accurate virtual screening of compounds against novel biological targets.