Abstract
Chromosomal conformation capture (3C) approaches, such as Hi-C, are increasingly implemented to reconstruct metagenome-assembled genomes (MAGs) from complex microbial communities,1but the precision of Hi-C-based binning has not been systematically studied. We evaluated the precision of Hi-C-based MAG reconstruction by implementing three metagenomic binning pipelines, ProxiMeta (MCMC-based), Bin3C (Infomap), and MetaTOR (Louvain community detection), to deconvolute MAGs from multiple Hi-C library replicates of a wastewater sample. We assessed precision by implementing the Jaccard distance metric to compare the derived MAGs. We found that the choice of the metagenomic pipeline plays a crucial role in the precision of Hi-C-based MAG reconstruction. All three pipelines produced highly reproduced MAGs when we reran the analysison the same Hi-C input, but across different Hi-C libraries, ProxiMeta achieved high reproducibility (median dissimilarity <0.05) at CheckM completeness of ≥65%, Bin3C at ≥80%, while MetaTOR did not reach this threshold at any completeness level. Cross-pipeline comparisons showed that ProxiMeta and Bin3C converged at a completeness level of ≥85% to produce highly similar MAGs, while MetaTOR never converged. These results indicate that variability in Hi-C-based MAG reconstruction is highly driven by Hi-C library preparation, as well as by the binning algorithm, and that CheckM completeness level plays a crucial role in MAG reproducibility.