PerturbQA

该数据集是对扰动实验的结构化推理的基准，重点关注对未见扰动条件下差异表达预测和变化方向的研究，同时包含基因集富集分析。该数据集还包括对最先进的机器学习和统计方法在建模扰动方面的评估，强调在此领域需要改进方法。任务涵盖了差异表达预测、变化方向预测以及基因集富集分析。

This dataset serves as a benchmark for structured reasoning in perturbation experiments, focusing on the prediction of differential expression and direction of expression changes under unseen perturbation conditions, and also includes gene set enrichment analysis. It further encompasses evaluations of state-of-the-art machine learning and statistical methods for perturbation modeling, emphasizing the critical need for improved methodologies in this field. The covered tasks include differential expression prediction, direction-of-change prediction, and gene set enrichment analysis.