Parsing Prokaryotic and Eukaryotic Microbial Genes Simultaneously from Metagenome by a Novel Analysis Workflow

Abstract Accurately predicting coding genes from metagenomic samples containing a high proportion of eukaryotic content is a current challenge. Noval reliable methods for the simultaneous prediction of prokaryotic and eukaryotic microbial genes could make contribution to the task. Our group evaluated gene prediction accuracy for various single organisms using MetaGeneMark and MetaEuk pipelines. Based upon the results, an innovative analysis workflow was devised. Our approach involves initially predicting eukaryotic genes using MetaEuk, followed by masking of predicted eukaryotic and partial prokaryotic genes via a Perl program. The subsequent prediction of remaining prokaryotic genes is conducted using MetaGeneMark or metaProdigal. By combining the advantages of both types of software, the new workflow achieved similar quantities and average lengths of eukaryotic genes compared to using MetaEuk alone, while the quantity of predicted prokaryotic genes and viral genes using the new workflow was 14–18% higher compared to using standalone prokaryotic gene prediction software alone. The new workflow effectively enables the rapid and precise retrieval of coding sequences from metagenomes.