Genomes from uncultivated prokaryotes: a comparison of metagenome-assembled and single-amplified genomes

Abstract

Background: Prokaryotes dominate the biosphere and regulate biogeochemical processes essential to all life. Yet, our knowledge about their biology is for the most part limited to the minority that has been successfully cultured. Molecular techniques now allow for obtaining genome sequences of uncultivated prokaryotic taxa, facilitating in-depth analyses that may ultimately improve our understanding of these key organisms.

Results: We compared results from two culture-independent strategies for recovering bacterial genomes: single-amplified genomes and metagenome-assembled genomes. Single-amplified genomes were obtained from samples collected at an offshore station in the Baltic Sea Proper and compared to previously obtained metagenome-assembled genomes from a time series at the same station. Among 16 single-amplified genomes analyzed, seven were found to match metagenome-assembled genomes, affiliated with a diverse set of taxa. Notably, genome pairs between the two approaches were nearly identical (average 99.51% sequence identity; range 98.77-99.84%) across overlapping regions (30-80% of each genome). Within matching pairs, the single-amplified genomes were consistently smaller and less complete, whereas the genetic functional profiles were maintained. For the metagenome-assembled genomes, only on average 3.6% of the bases were estimated to be missing from the genomes due to wrongly binned contigs.

Conclusions: The strong agreement between the single-amplified and metagenome-assembled genomes emphasizes that both methods generate accurate genome information from uncultivated bacteria. Importantly, this implies that the research questions and the available resources are allowed to determine the selection of genomics approach for microbiome studies.

Keywords: Binning; Metagenome-assembled genomes; Metagenomics; Single-amplified genomes; Single-cell genomics.

Fig. 1

Gene homolog presence per genome cluster. Presence of gene homologs for each genome cluster by graphs produced by Anvi’o. Each horizontal bar represents one genome, where blue bars are single-amplified genomes and black and grey bars are metagenome-assembled genomes. Each vertical bar corresponds to one gene homolog where a dark vertical bar indicates presence of the gene homologs and a lighter vertical bar indicates absence. The gene homologs are aligned between genomes within each genome cluster. The numbers assigned to the genome clusters corresponds to the original MAG BACLs used in [23]

Fig. 2

Distribution of functional categories in SAGs and MAGs. a Distribution of broad COG categories in the different genome clusters for MAGs and SAGs. The X-axis shows genomes grouped and ordered according to genome clusters. The Y-axis shows the percentage of genes in COG categories in each genome. b Non-metric multidimensional scaling (NMDS) plot based on counts of COG categories in the SAGs and MAGs

Fig. 3

Distribution of SAG reads mapped against metagenome assemblies. Boxplot of the distribution of SAG reads mapped against the corresponding metagenome assemblies where each individual data point is jittered on top of each box. All reads for each SAG was mapped against the assembly associated with each matching MAG and thus positioned in exactly one out of these four categories. Only contigs longer than 1 kb were included in the binning, which is the reason to use it as a divider here