Paper List
-
PanFoMa: A Lightweight Foundation Model and Benchmark for Pan-Cancer
This paper addresses the dual challenge of achieving computational efficiency without sacrificing accuracy in whole-transcriptome single-cell represen...
-
Beyond Bayesian Inference: The Correlation Integral Likelihood Framework and Gradient Flow Methods for Deterministic Sampling
This paper addresses the core challenge of calibrating complex biological models (e.g., PDEs, agent-based models) with incomplete, noisy, or heterogen...
-
Contrastive Deep Learning for Variant Detection in Wastewater Genomic Sequencing
This paper addresses the core challenge of detecting viral variants in wastewater sequencing data without reference genomes or labeled annotations, ov...
-
SpikGPT: A High-Accuracy and Interpretable Spiking Attention Framework for Single-Cell Annotation
This paper addresses the core challenge of robust single-cell annotation across heterogeneous datasets with batch effects and the critical need to ide...
-
Unlocking hidden biomolecular conformational landscapes in diffusion models at inference time
This paper addresses the core challenge of efficiently and accurately sampling the conformational landscape of biomolecules from diffusion-based struc...
-
Learning From Limited Data and Feedback for Cell Culture Process Monitoring: A Comparative Study
This paper addresses the core challenge of developing accurate real-time bioprocess monitoring soft sensors under severe data constraints: limited his...
-
Cell-cell communication inference and analysis: biological mechanisms, computational approaches, and future opportunities
This review addresses the critical need for a systematic framework to navigate the rapidly expanding landscape of computational methods for inferring ...
-
Generating a Contact Matrix for Aged Care Settings in Australia: an agent-based model study
This study addresses the critical gap in understanding heterogeneous contact patterns within aged care facilities, where existing population-level con...
Mechanistic Interpretability of Antibody Language Models Using SAEs
Department of Statistics, University of Oxford, UK | Reticular, San Francisco, USA | Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology (MIT), Cambridge, MA, USA
The 30-Second View
IN SHORT: This work addresses the core challenge of achieving both interpretability and controllable generation in domain-specific protein language models, specifically for antibody design.
Innovation (TL;DR)
- Methodology First application of Sparse Autoencoders (SAEs) to interrogate autoregressive antibody-specific language models (p-IgGen), moving beyond general protein language models.
- Methodology Systematic comparison reveals a key trade-off: TopK SAEs yield highly interpretable, monosemantic features (e.g., for CDR identity with validation accuracy 0.99) but lack causal steerability, while Ordered SAEs provide reliable generative control at the cost of interpretability.
- Biology Identifies and validates antibody-specific, biologically meaningful latent features, such as CDR identity and germline gene identity (e.g., IGHJ4 prediction with F1 macro score of 0.93), demonstrating the model's learning of immunologically relevant concepts.
Key conclusions
- TopK SAEs effectively compress and preserve biological information (CDR identity prediction accuracy 0.99 vs. 0.98 for raw neurons) and yield sparse, interpretable activation patterns localized to specific regions (e.g., CDRH3), overcoming neuron polysemanticity.
- High feature-concept correlation (e.g., F1 > 0.5 for IGHJ4 latents) does not guarantee causal steerability; steering on TopK-identified IGHJ4 features failed to consistently increase IGHJ4 proportions in generated sequences.
- Ordered SAEs, with their enforced hierarchical latent structure (via per-index nested grouping and decreasing truncation weights), successfully identify features that enable predictable generative steering, albeit with more complex activation patterns.
Abstract: Sparse autoencoders (SAEs) are a mechanistic interpretability technique that have been used to provide insight into learned concepts within large protein language models. Here, we employ TopK and Ordered SAEs to investigate an autoregressive antibody language model, p-IgGen, and steer its generation. We show that TopK SAEs can reveal biologically meaningful latent features, but high feature–concept correlation does not guarantee causal control over generation. In contrast, Ordered SAEs impose an hierarchical structure that reliably identifies steerable features, but at the expense of more complex and less interpretable activation patterns. These findings advance the mecahnistic interpretability of domain-specific protein language models and suggest that, while TopK SAEs suffice for mapping latent features to concepts, Ordered SAEs are preferable when precise generative steering is required.