PDBEurope/protein_chain_conformational_states

该数据集是一个手动整理的蛋白质链构象状态数据集，包含了开放-关闭单体对以及中间状态，这些状态由条目的作者定义。数据集通过查询PDBe数据库，筛选出在UniProt段中具有100%序列同一性的开放和关闭构象的PDBe条目，并进行了手动整理。整理过程包括清理不相关的条目、手动调查每个PDBe条目的对应出版物、添加遗漏的条目、记录中间构象以及移除以多聚体形式解析的条目。数据集还通过PDBe-KB的现有聚类算法进行了交叉引用，以评估当前构象聚类的成功。

原始信息汇总

数据集概述

数据集名称

Curated dataset of protein chain conformational states

数据集标签

• Structural biology
• Bioinformatics
• Machine learning
• Conformation
• Conformational state
• Monomeric
• Training data
• Benchmark
• Manually curated

数据集大小

10K<n<100K

任务类别

• feature-extraction

语言

• en

许可证

cc-by-4.0

数据集文件

benchmarking_monomeric_open_closed_conformers.csv

数据集列描述

• UNP_ACC | UniProt accession code
• UNP_START | Start of UniProt sequence for given PDBe entries
• UNP_END | End of UniProt sequence for given PDBe entries
• PDBe_ID | Protein Data Bank code
• CHAIN_ID | Author declared chain ID (char)
• label_asym_id | Programmatically assigned chain ID (char)
• CONFORMER_ID | Unique code for PDBe entries with distinct conformation, corresponding to a given UniProt accession
• CONFORMER_DESCR | Short description of conformation, based on depositors assessment of the protein/conformation
• LIT_CONFIRMED | True/false value based on whether a publication (scientific literature) was available for manually curating clusters. NB: Clusters with 0 in this field should be used with caution.
• ALT_CONFORMER_ID | Where the publication for a structure is currently outstanding, an executive decision on the conformation classification is made. Where the literature is not explicit on the features of a given conformation, the second most suitable CONFORMER_ID is provided in this column. Blank cells have no other likely conformation assignmnt and are therefore the same as in CONFORMER_ID.
• ALT_CONFORMER_DESCR | Description for conformation in alternative conformation ID.

数据集整理过程

1. 通过PDBe的Oracle数据库查询，找到在TITLE字段中同时具有open和closed构象的PDBe条目，且序列完全相同。
2. 清理结果，移除TITLE字段中包含open或close但不指代构象的条目。
3. 手动检查剩余条目，并根据相应出版物进行整理：
   • 添加作者提交的、在原始搜索中因缺少open或close字段而遗漏的条目。
   • 记录作者报告的中间构象。
   • 移除以单体形式提交但解决为多聚体复合物的条目。
4. 将PDBe条目按作者声明的构象进行聚类，并与PDBe-KB的现有聚类算法进行交叉验证。

整理流程图

数据集总结图