Assembly-free short-read metagenomic maximum growth rate prediction

Abstract Microbial maximum growth rates are key parameters in microbially explicit soil biogeochemical models that can also be estimated directly from genomes and metagenomes, therefore enabling model-data synthesis between coarse-grained models and high-resolution sequencing data. A persistent challenge for predicting community-level growth phenotypes in soil metagenomes is that current maximum growth rate prediction models require complete gene sequences from assembled contigs, but the complexity of soil metagenomes often precludes high quality assembly. Here, we develop a set of gene length specific models that are capable of predicting community average maximum growth rates directly from genes inferred from reads, avoiding the need for an assembly step. These models have been implemented in the open source gRodon R package for maximum growth rate prediction.