Paper List
-
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...
-
Personalized optimization of pediatric HD-tDCS for dose consistency and target engagement
This paper addresses the critical limitation of one-size-fits-all HD-tDCS protocols in pediatric populations by developing a personalized optimization...
-
Realistic Transition Paths for Large Biomolecular Systems: A Langevin Bridge Approach
This paper addresses the core challenge of generating physically realistic and computationally efficient transition paths between distinct protein con...
-
Consistent Synthetic Sequences Unlock Structural Diversity in Fully Atomistic De Novo Protein Design
This paper addresses the core pain point of low sequence-structure alignment in existing synthetic datasets (e.g., AFDB), which severely limits the pe...
-
MoRSAIK: Sequence Motif Reactor Simulation, Analysis and Inference Kit in Python
This work addresses the computational bottleneck in simulating prebiotic RNA reactor dynamics by developing a Python package that tracks sequence moti...
-
On the Approximation of Phylogenetic Distance Functions by Artificial Neural Networks
This paper addresses the core challenge of developing computationally efficient and scalable neural network architectures that can learn accurate phyl...
-
EcoCast: A Spatio-Temporal Model for Continual Biodiversity and Climate Risk Forecasting
This paper addresses the critical bottleneck in conservation: the lack of timely, high-resolution, near-term forecasts of species distribution shifts ...
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
30秒速读
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.
核心创新
- 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.
主要结论
- 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.
摘要: 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.