proteinglm/solubility_prediction

溶解度预测数据集用于蛋白质溶解度的二元分类任务，包含蛋白质序列字符串和表示溶解度的整数标签。数据集分为训练集和测试集，分别包含62,478和6,942个实例。数据来源于DeepSol，并遵循Apache-2.0许可。

This solubility prediction task involves a binary classification of a heterogenous set of proteins, assessing them as either soluble or insoluble. The solubility metric is a crucial design parameter in ensuring protein efficacy, with particular relevance in the pharmaceutical domain. The dataset includes two main parts: protein sequence (seq) and label (label), where the label indicates the solubility of the protein sequence. The dataset is divided into training and test sets, containing 62,478 and 6,942 instances respectively. The source data is from DeepSol and has been normalized to ensure robust and unbiased evaluation. The dataset is released under the Apache-2.0 license.