AUTHOR=Lu Yuntao , Li Qi , Li Tao TITLE=PPA-GCN: A Efficient GCN Framework for Prokaryotic Pathways Assignment JOURNAL=Frontiers in Genetics VOLUME=13 YEAR=2022 URL=https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2022.839453 DOI=10.3389/fgene.2022.839453 ISSN=1664-8021 ABSTRACT=
With the rapid development of sequencing technology, completed genomes of microbes have explosively emerged. For a newly sequenced prokaryotic genome, gene functional annotation and metabolism pathway assignment are important foundations for all subsequent research work. However, the assignment rate for gene metabolism pathways is lower than 48% on the whole. It is even lower for newly sequenced prokaryotic genomes, which has become a bottleneck for subsequent research. Thus, the development of a high-precision metabolic pathway assignment framework is urgently needed. Here, we developed PPA-GCN, a prokaryotic pathways assignment framework based on graph convolutional network, to assist functional pathway assignments using KEGG information and genomic characteristics. In the framework, genomic gene synteny information was used to construct a network, and ideas of self-supervised learning were inspired to enhance the frameworkâs learning ability. Our framework is applicable to the genera of microbe with sufficient whole genome sequences. To evaluate the assignment rate, genomes from three different genera (