EWCL–MobiDB Benchmarks 2025: DisProt, IDEAL, Merge, UniProt, and Binding Snapshots
收藏资源简介:
This dataset provides per-residue and per-protein benchmark tables for EWCLv1 (Entropy-Weighted Collapse Likelihood) evaluated against the MobiDB curated disorder references — DisProt, IDEAL, Merge, and UniProt — as well as the derived Binding Disorder–Disorder (D→D) snapshot. EWCL quantifies local conformational entropy and thermodynamic confidence directly from sequence. Each archive includes ROC and PR plots, per-protein metrics, and summary JSONs used in the EWCL-2025 study. These benchmarks complement the EWCL–CAID Benchmarks 2025 dataset (Zenodo DOI: 10.5281/zenodo.17352525), which provides CAID-aligned disorder, linker, and binding evaluations. Contents EWCL_MOBIDB_2025_Disprot.zip — DisProt reference (full curated set) EWCL_MOBIDB_2025_IDEAL_excl_Disprot.zip — IDEAL benchmark excluding DisProt overlap EWCL_MOBIDB_2025_Merge_excl_Disprot.zip — MobiDB Merge benchmark excluding DisProt overlap EWCL_MOBIDB_2025_UniProt_excl_Disprot.zip — UniProt benchmark excluding DisProt overlap EWCL_MOBIDB_2025_Binding_DD.zip — Non-homology Binding Disorder–Disorder (D→D priority) snapshot Each ZIP contains: data/ (residue- and protein-level CSVs), figs/ (ROC and PR curves), and meta/summary.json (micro/macro AUCs, F1, prevalence)
本数据集提供了针对EWCLv1(Entropy-Weighted Collapse Likelihood,熵加权折叠似然度)的残基水平与蛋白质水平基准测试表,其评估基准基于MobiDB经人工注释整理的无序蛋白质参考数据集——DisProt、IDEAL、Merge及UniProt——以及衍生的结合型无序-无序(D→D)快照。 EWCL可直接从氨基酸序列量化局部构象熵与热力学置信度。 每个归档文件均包含受试者工作特征(ROC)曲线、精确率-召回率(PR)曲线、蛋白质水平评估指标以及用于EWCL-2025研究的总结JSON文件。 本基准测试数据集是对EWCL–CAID 2025基准数据集(Zenodo DOI: 10.5281/zenodo.17352525)的补充,后者提供了与CAID对齐的无序蛋白质、连接区域以及结合特性评估结果。 数据集内容 EWCL_MOBIDB_2025_Disprot.zip — DisProt参考数据集(完整经人工注释整理的集合) EWCL_MOBIDB_2025_IDEAL_excl_Disprot.zip — 去除与DisProt重叠序列的IDEAL基准测试集 EWCL_MOBIDB_2025_Merge_excl_Disprot.zip — 去除与DisProt重叠序列的MobiDB Merge基准测试集 EWCL_MOBIDB_2025_UniProt_excl_Disprot.zip — 去除与DisProt重叠序列的UniProt基准测试集 EWCL_MOBIDB_2025_Binding_DD.zip — 非同源结合型无序-无序(D→D优先)快照数据集 每个ZIP压缩包均包含以下内容: data/(残基水平与蛋白质水平的CSV格式数据)、 figs/(受试者工作特征曲线与精确率-召回率曲线)、 以及meta/summary.json(存储微观/宏观AUC值、F1值与样本患病率)



