Accurate Reconstruction of Microbial Strains from Metagenomic

Exploring the genetic diversity of microbes within the environment through metagenomic 5 sequencing first requires classifying these reads into taxonomic groups. Current methods compare these 6 sequencing data with existing biased and limited reference databases. Several recent evaluation studies 7 demonstrate that current methods either lack sufficient sensitivity for species-level assignments or suffer 8 from false positives, overestimating the number of species in the metagenome. Both are especially prob9 lematic for the identification of low-abundance microbial species, e. g. detecting pathogens in ancient 10 metagenomic samples. We present a new method, SPARSE, which improves taxonomic assignments 11 of metagenomic reads. SPARSE balances existing biased reference databases by grouping reference 12 genomes into similarity-based hierarchical clusters, implemented as an efficient incremental data struc13 ture. SPARSE assigns reads to these clusters using a probabilistic model, which specifically penalizes 14 non-specific mappings of reads from unknown sources and hence reduces false-positive assignments. 15 Our evaluation on simulated datasets from two recent evaluation studies demonstrated the improved 16 precision of SPARSE in comparison to other methods for species-level classification. In a third simula17 tion, our method successfully differentiated multiple co-existing Escherichia coli strains from the same 18 sample. In real archaeological datasets, SPARSE identified ancient pathogens with ≤ 0.02% abundance, 19 consistent with published findings that required additional sequencing data. In these datasets, other 20 methods either missed targeted pathogens or reported non-existent ones. 21 SPARSE and all evaluation scripts are available at https://github.com/zheminzhou/SPARSE. 22