proteinglm/contact_prediction_binary

接触图预测数据集旨在确定两个残基是否在接触范围内，基于它们的距离是否小于8埃。该任务是早期Alphafold版本中结构预测的重要组成部分。数据集包含蛋白质序列和接触标签，标签表示残基对是否接触。数据集分为训练、验证和测试三个部分，分别包含12,041、1,505和1,505个实例。数据集的初始数据来源于trRosetta数据集，并遵循Apache-2.0许可证。

The Contact Map Prediction Dataset aims to determine whether two residues, $i$ and $j$, are in contact based on their distance (less than 8 Angstrom). This task is an important part of the early Alphafold version for structural prediction. The dataset contains protein sequences and contact labels, with each instance including a protein sequence string and a contact label sequence. The dataset is divided into train, validation, and test splits, containing 12,041, 1,505, and 1,505 instances respectively. The features of the dataset include protein sequences and contact labels, with average lengths of 249 and 1,500 respectively. The dataset is based on the trRosetta dataset and is released under the Apache-2.0 license.