Building better microbial infection models: a call to do the "field" experiments.

Abstract

Experimental model systems are essential in microbiology. However, models are often described as "biologically relevant" without a clear explanation of what that means or how relevance was established. Here, I argue that the missing piece in model development is benchmarking. New technologies and increasingly elaborate model systems can be powerful, but they do not guarantee that a model better represents the environment it is meant to reproduce. The guarantee that is implied cannot be made explicit until these models are evaluated against measurements of microbial behavior and function made directly in those environments. These "field" experiments are often messy, expensive, low throughput, and technically challenging, but provide the benchmarks needed to determine what a model captures, what it misses, and which questions it can tackle. While chemical measurements and quantification of physical features can guide model construction, the most important readout is the behavior of the microbes in the natural environment.