Relational completion based non-negative matrix factorization for predicting metabolite-disease associations

作者：

Highlights：

• We develop a computational model called RCNMF to predict the metabolite-disease association.

• The molecular structure internal information of metabolite is used to compute the similarity of metabolites.

• Our method achieves good performance by reducing the sparsity of the association between metabolites and disease.

摘要

•We develop a computational model called RCNMF to predict the metabolite-disease association.•The molecular structure internal information of metabolite is used to compute the similarity of metabolites.•Our method achieves good performance by reducing the sparsity of the association between metabolites and disease.

论文关键词：Metabolite-disease associations,Non-negative matrix factorization,Molecular fingerprint similarity of metabolite

论文评审过程：Received 27 October 2019, Revised 8 June 2020, Accepted 8 July 2020, Available online 13 July 2020, Version of Record 15 July 2020.

论文官网地址：https://doi.org/10.1016/j.knosys.2020.106238