Metagenome Assembly Validation: Which Metagenome Contigs are Bona Fide?

In the metagenomics, long metagenome contigs can either improve metagenome gene prediction or metagenome sequence binning. Moreover, metagenome contigs can also make gene function annotation more accurate because they provide a lot of genome context information. Because of repetitive sequences of either intra-genomes or inter-genomes, metagenome contigs are probably wrongly assembled. Therefore, it is essential to develop a method to validate metagenome contigs. Here, we propose a computational method to validate metagenome contigs. After realigning raw sequencing reads onto one contig, we first compute a contig-ECDF (empirical cumulative probability distribution functions) and its corresponding reference using a computational simulation-based method. Because a reference of the contig-ECDF is changeless given some parameters, we use the distinction between them to check whether or not a contig is bona fide. The less the distinction is, the more likely a contig is bona fide. For wrongly assembled metagenome contigs, using simulated metagenome datasets, our method was shown to have a good capacity to identify them. After applying the method to a real metagenome dataset, which was sequenced from an in vitro-simulated microbial community with known constituted genomes, we showed that our method had a strong ability to identify bona fide contigs, and further demonstrated that small distinctions between contig-ECDFs and their references were significantly correlated with bona fide contigs. A computational method is developed to validate metagenome contigs. For each metagenome contig, our method gives it a score, and the smaller the score is, the more likely a contig is bona fide. After validation using both simulated and real datasets, our method was shown to have good performances.