OpenMed/agab-db

AgAb DB是一个全面的抗体-抗原相互作用数据集，专为计算生物学和治疗设计而构建。该数据集汇集了来自多个来源的抗体-抗原结合数据，包含超过120万对抗体-抗原结合亲和力测量记录。数据集包含完整的重链和轻链序列、抗原序列、结合亲和力值以及数据质量等级等核心字段。此外，还提供了丰富的元数据，如抗体类型（纳米抗体、scFv等）、抗原名称、PDB结构ID、UniProt编号等。数据集特别关注人类健康相关领域，包括传染病、癌症和自身免疫性疾病，并包含多种抗原类型（如病毒蛋白、细菌抗原、癌症标志物等）。数据质量分为高、中、低三个置信度等级，其中51%为极高置信度数据。该数据集主要用于训练计算免疫学和抗体工程中的机器学习模型，支持抗体语言模型训练、结合亲和力预测、治疗设计等多种应用场景。

AgAb DB is a comprehensive collection of antibody-antigen interaction data for computational biology and therapeutic design. The dataset aggregates antibody-antigen binding data from multiple sources, containing over 1.2 million antibody-antigen pairs with binding affinity measurements. It includes core fields such as complete heavy and light chain sequences, antigen sequences, binding affinity values, and data quality levels. Additionally, it provides rich metadata including antibody types (nanobodies, scFvs, etc.), antigen names, PDB structure IDs, UniProt accessions, and more. The dataset focuses on human health-related areas including infectious diseases, cancer, and autoimmune conditions, and encompasses diverse antigen types (viral proteins, bacterial antigens, cancer markers, etc.). Data quality is categorized into three confidence levels (very_high, high, medium), with 51% being very high confidence. Primarily designed for training machine learning models in computational immunology and antibody engineering, the dataset supports various applications such as antibody language model training, binding affinity prediction, and therapeutic design.