We introduce the Fitness Landscape for Antibodies (FLAb2), the largest public benchmark for therapeutic antibody design, comprising developability data for over 4 million antibodies across 32 studies and covering 7 properties: thermostability, expression, aggregation, binding affinity, pharmacokinetics, polyreactivity, and immunogenicity. Evaluating 30 AI and biophysical models, we find that most (80%) fail to produce statistically significant correlations. No model performs well across all properties. Some pretrained models (IgLM, ProGen2, Chai-1, ESM2, ISM, and IgFold) show significant zero-shot correlations for select properties. Fine-tuning with ≥102 points helps for thermostability, aggregation, and binding though polyreactivity and pharmacokinetics lack sufficient data. With enough data (103 points), a fine-tuned one-hot encoding model matches billion-parameter models. Training data matters more than architecture, and intrinsic properties (thermostability) are predicted more readily than extrinsic ones (immunogenicity and pharmacokinetics). Partial correlation analysis shows that germline distance accounts for ∼37% of language models’ apparent predictive power.
Digital pathology has become a vital component of modern oncology. Advanced imaging and staining technologies have accelerated the integration of artificial intelligence (AI). While these methods enable automated clinical analysis, their lack of interpretability can become a bottleneck for trustworthy deployment and regulatory compliance. This article provides a comprehensive review of interpretable AI methods and applications in digital pathology over the past decade (N = 276, 2015–2026). Unlike previous contributions, we use a methodology-centric perspective by offering a systematic taxonomy of interpretable AI methods, grouped into interpretable-by-design and post-hoc techniques. Within each category, they are further analyzed by their utility in key digital pathology problems and applications. A complementary technical handbook on interpretable AI is provided for readers seeking a deeper understanding of the methods themselves. By reviewing the current state, highlighting limitations, and outlining potential future directions, this review aims to accelerate the trustworthy adoption of AI in real-world clinical settings.
