Data underlying the publication: "GONNECT: Coupling Biological Systems to Neural Networks for Improved Model Interpretability"

This dataset contains all processed data required to reproduce the results of the GONNECT paper. In this work, we couple the structure of a neural network model to biological prior information, to gain interpretable activations in the neural network's hidden layers (see preprint/publication for more information). The data presented here includes processed gene expression data from The Cancer Genome Atlas (TCGA) data that is used as input data for the model, and both the raw and processed Gene Ontology (GO, https://geneontology.org/) knowledge base, from which the structure of the neural networks in this study is derived. There are also some miscellaneous files used to link genes to proteins. All data used in this study is publically available.<br>Code corresponding to the paper can be found here: https://github.com/DelftBioinformaticsLab/GONNECT

本数据集包含复现GONNECT论文研究成果所需的全部预处理数据。本研究将神经网络模型的结构与生物学先验信息相结合，以在神经网络的隐藏层中获得可解释的激活结果（详细信息可参见预印本或已发表论文）。本次公开的数据包含两类核心内容：其一为源自癌症基因组图谱（The Cancer Genome Atlas, TCGA）的预处理基因表达数据，该数据将作为模型的输入数据；其二为原始与经过预处理的基因本体（Gene Ontology, GO, https://geneontology.org/）知识库，本研究中的神经网络结构即源自该知识库。此外还包含若干用于关联基因与蛋白质的辅助文件。本研究涉及的全部数据均为公开可获取的。对应本论文的研究代码可通过以下链接获取：https://github.com/DelftBioinformaticsLab/GONNECT