生物学定量方法：2025–2026 年度综述

An annual survey of the q-bio.QM arXiv category, covering research published from September 2025 – August 2026. Built on a close-reading set of 120 papers and organised into 44 chapters, it maps the year by the field's research sub-directions.

Book framework (share of the year's output):

• AI-Driven Molecular Property and Clinical Prediction (16.6%) — Focuses on deep learning and multimodal fusion techniques for predicting molecular properties, disease outcomes, and clinical survival metrics.
• Automated Image Analysis and Diagnostic Detection (12.8%) — Addresses automated detection, segmentation, and classification tasks in medical imaging, microscopy, and biological data analysis.
• Generative AI for Drug Discovery and Molecular Design (12.4%) — Explores diffusion models, generative AI, and agentic frameworks for generating novel molecules and optimizing drug discovery pipelines.
• Protein Language Models and Structural Generation (10.9%) — Utilizes protein language models and autoregressive methods for protein design, sequence generation, and structural dynamics modeling.
• Foundation Models in Biomedical Imaging and Biology (10.0%) — Investigates the application and training of large foundation models for biomedical imaging, virtual staining, and biological sequence understanding.
• Physics-Informed and Graph Neural Networks (9.6%) — Integrates physics-informed constraints and graph neural networks to model biological systems, tissue properties, and dynamic processes.
• Single-Cell and Spatial Multi-Omics Modeling (8.4%) — Develops computational frameworks for integrating single-cell multi-omics data, modeling cell perturbations, and analyzing spatial transcriptomics.
• Automated Biological Data Processing and Identification (7.3%) — Focuses on automated pipelines for identifying biological entities, detecting diseases, and processing large-scale biological datasets.
• Deep Learning Frameworks for Disease Classification (7.1%) — Applies deep learning and machine learning frameworks to classify diseases, detect biomarkers, and analyze clinical data with robustness.
• Multimodal Data Integration and Statistical Analysis (5.0%) — Employs statistical methods and multimodal integration techniques for analyzing complex biological data, including fNIRS, genomic, and clinical records.

Each sub-direction is covered across three chapters — how it framed its questions, the strength of its key findings, and its open debates and frontiers — so the reader gets a traceable picture of what this part of the field actually produced over the year.