Biases in genome reconstruction from metagenomic data

• DOI: 10.7717/peerj.10119
• Publication Date: 2020-10-30T09:58:06Z
• Authors: William C. Nelson; Benjamin J. Tully; Jennifer M. Mobbe...
• Keywords: RefSeq

Background Advances in sequencing, assembly, and assortment of contigs into species-specific bins has enabled the reconstruction genomes from metagenomic data (MAGs). Though a powerful technique, it is difficult to determine whether assembly binning techniques are accurate when applied environmental metagenomes due lack complete reference genome sequences against which check resulting MAGs. Methods We compared MAGs derived an enrichment culture containing ~20 organisms 10 isolated culture. Factors commonly considered software—nucleotide composition sequence repetitiveness—were calculated for both correctly binned not-binned regions. This direct comparison revealed biases characteristics gene content Additionally, three public sets representing reconstructed Tara Oceans was set representative available through NCBI RefSeq verify that identified were observable more complex using contemporary software packages. Results Repeat frequently not processes, as regions with variant nucleotide composition. Genes encoded on strongly biased towards ribosomal RNAs, transfer mobile element functions genes unknown function. Our results support robust process suggest reconstructions determined be >90% likely effectively represent organismal function; however, population-level genotypic heterogeneity natural populations, such uneven distribution plasmids, can lead incorrect inferences.