AUTHOR=Zhang Yi , Wang Yu , Li Xin , Liu Yarong , Chen Min 

TITLE=Identifying lncRNA–disease association based on GAT multiple-operator aggregation and inductive matrix completion

JOURNAL=Frontiers in Genetics

VOLUME=Volume 13 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2022.1029300

DOI=10.3389/fgene.2022.1029300

ISSN=1664-8021

ABSTRACT=Computable models as a fundamental candidate for traditional biological experiments have been applied in inferring lncRNA-disease associations (LDA) for many years, without time-consuming and laborious limitations. However, sparsity inherently existing in known heterogeneous bio-data is an obstacle of  computable models to improve prediction accuracy further. Therefore, a new computational model that was composed of Multiple Mechanisms for LncRNA-Disease Association prediction (MM-LDA) was proposed, based on the fusion of graph attention network (GAT) and inductive matrix completion (IMC). MM-LDA has two key steps to improve prediction accuracy: First step, a multiple-operator aggregation was designed in the n-heads attention mechanism of GAT. With this step, features of lncRNA nodes and disease nodes were enhanced. Second step, IMC was introduced into the enhanced node features obtained in the first step, then the LDA network was reconstructed to solve the cold start problem when data deficiency of the entire row or column happened in known association matrix. Our MM-LDA achieved following progress: First, using Adam optimizer that adaptatively adjusted the model learning rate could increase convergent speed and not fall into local optima as well. Second, more excellent predictive ability was achieved against other similar models (with AUC value of 0.9395 and AUPR value of 0.8057 obtained from 5-fold-cross-validation). Third, 6.45% lower time cost was consumed against the advanced model GAMCLDA. In short, our MM-LDA achieved the more comprehensive prediction performance in terms of prediction accuracy and time cost.