Generation and application of pseudo-long reads for metagenome assembly

Background Metagenomic assembly using high-throughput sequencing data is a powerful method to construct microbial genomes in environmental samples without cultivation. However, metagenomic assembly, especially when only short reads are available, is a complex and challenging task because mixed genomes of multiple microorganisms constitute the metagenome. Although long read sequencing technologies have been developed and have begun to be used for metagenomic assembly, many metagenomic studies have been performed based on short reads because the generation of long reads requires higher sequencing cost than short reads. Results In this study, we present a new method called PLR-GEN. It creates pseudo-long reads from metagenomic short reads based on given reference genome sequences by considering small sequence variations existing in individual genomes of the same or different species. When applied to a mock community data set in the Human Microbiome Project, PLR-GEN dramatically extended short reads in length of 101 bp to pseudo-long reads with N50 of 33 Kbp and 0.4% error rate. The use of these pseudo-long reads generated by PLR-GEN resulted in an obvious improvement of metagenomic assembly in terms of the number of sequences, assembly contiguity, and prediction of species and genes. Conclusions PLR-GEN can be used to generate artificial long read sequences without spending extra sequencing cost, thus aiding various studies using metagenomes.

Automated summary

A new method called PLR-GEN is proposed to generate pseudo-long reads from metagenomic short reads by considering small sequence variations existing in individual genomes. The use of these pseudo-long reads significantly improves metagenomic assembly in terms of sequence number, assembly contiguity, and prediction of species and genes.