MetaPore-C: using chromatin conformation capture and long nanopore reads for metagenomic analysis

Chromatin conformation capture is a method used to investigate interactions between genomic loci that are not adjacent in the primary sequence. When applied to metagenomic mixtures, as well as simplifying assembly the technique provides a way to associate plasmids with their host genomes. It is typical to use restriction digestion to fragment cross-linked DNA before proximity ligation, but this step is influenced by the base composition of the genomes present, which is not always known in advance. Our protocol avoids digestion by using bead beating to simultaneously lyse the cells and to fragment the DNA (Fig. 1a). To produce contact information, MetaPore-C reads are first aligned to a collection of chromosomal and extra-chromosomal reference sequences using BWA-SW. Aligned reads are filtered to retain the minimal collection of alignments that traverse the majority of the read. The reference genomes are then divided into equally sized bins and each aligned segment of the MetaPore-C read is assigned a bin. Finally, the total number of bin-to-bin contacts is calculated from all reads and visualised in a contact map. Extra-chromosomal elements can be assigned to their host by determining which chromosome(s) share the most contacts with the element (Fig. 1b). This approach allowed us to assign plasmids to the correct host in a mixture of two bacteria (Fig. 1c).

We generated MetaPore-C data on a probiotic food supplement, which contained 15 known bacterial strains (Fig. 2a) and created contact maps for the bacterial chromosomes and plasmids within this sample (Figs. 2b and 2c). The plot of average nucleotide identity (Fig. 2d) reveals a low level of spurious interaction between species, most probably due to mapping ambiguities. Future work will involve the use of more fine-grained approaches to suppress this background signal. Figs. 2e and 2f summarise the contacts for each bacterial chromosome. Plasmids were associated to the expected host genomes and we identified intra-chromosomal interactions, which are valuable for binning and hence assembly.

To see if we could correctly associate plasmids even in extremely similar bacterial host genomes, we created a MetaPore-C contact map for a mock community with 6 bacterial strains and their plasmids (Fig. 3a). The community includes two E. coli strains, plus S. enterica, which shares >80% ANI with E. coli. The long regions of high contact density reveal the plasmid-genome contacts. There are small amounts of spurious, between-species interactions, due to ambiguous mapping between highly homologous regions. When the proportion of contacts is plotted for each possible plasmid-genome interaction, in every case the overwhelming majority of contacts is with the correct host genome (Fig. 3b).