Influence of sequencing depth on bacterial classification and abundance in bacterial communities

Microbial diversity is the most abundant form of life. Next Generation Sequencing technologies provide the capacity to study complex bacterial communities, in which the depth and the bioinformatic tools can influence the results. In this work we explored two different protocols for bacterial classification and abundance evaluation, using 10 bacterial genomes in a simulated sample at different sequencing. Protocol A consisted of metagenome assembly with Megahit and Ray Meta and taxonomic classification with Kraken2 and Centrifuge. In protocol B only taxonomic classification. In both protocols, rarefaction, relative abundance and beta diversity were analyzed. In the protocol A, Megahit had a mean contig length of 1,128 and Ray Meta de 8,893 nucleotides. The number of species correctly classified in all depth assays were 6 out of 10 for protocol A, and 9 out of 10 using protocol B. The rarefaction analysis showed an overestimation of the number of species in almost all assays regardless of the protocol, and the beta diversity analysis results indicated significant differences in all comparisons. Protocol A was more efficient for diversity analysis, while protocol B estimated a more precise relative abundance. Our results do not allow us to suggest an optimal sequencing depth at specie level.