Devising a General Microbial Model Evaluation Framework for Microbiology Research

An article published by the American Society for Microbiology has outlined a general quantitative framework to determine the accuracy of in vitro and in vivo models used in microbiology research. The team evaluated the efficacy of their model using the bacterium Pseudomonas aeruginosa.

Pseudomonas aeruginosaImage Credits: Jun MT /

The models used to assemble the quantitative framework was successful in capturing several facets of the bacterium's infection physiology. Moreover, the group identified what model underperformed in functional categories, demonstrating where the model failed to capture this information.

The models included in a quantitative framework are hoped to provide researchers with an accurate method of selecting the optimal laboratory model for use in microbiological studies; one that accounts for the line of scientific inquiry.

There is typically a range of laboratory models used to study infections. The selection of models is largely based on ad hoc rationalizations or the researcher’s intuition. However, the assessment of a model’s performance is currently lacking due to a lack of a formalized framework.

The team sought to address the issue by proposing an RNA sequencing (RNA-seq) based framework to evaluate human infection models. RNA-Seq reveals pathogenic gene expression in an unperturbed manner - conditions that are difficult to achieve in vitro.

The motivation for this study was the study of Pseudomonas aeruginosa infection of the lungs of cystic fibrosis patients. The viscous mucus in patients provides an optimal environment for the colonization by P. aeruginosa. This bacterium relies on mucus as a source of carbon in energy, resulting in its growth at high density.

Although models can be rationalized based on their ability to recapitulate the physiological conditions in humans, for the study of P. aeruginosa infection, the relevance of these models remains to be assessed and many do not capture all characteristics of P. aeruginosa infection.

Study design

The group analyzed P. aeruginosa transcriptomes from 20 human CF sputum samples taken after expectoration. Specifically, they compared the transcriptomes taken from the sputum to those of a reference strain PAO1.

To assess the model accuracy, the group implemented a computational framework based on transcriptomic data, together with this data derived from CF infection models. This was used to gauge the performance of the model’s performance.

This was applied to the following experimental models:

  • Lysogeny broth (LB)
  • Mouse pneumonia model
  • Synthetic sputum medium (SCFM2)
  • in vitro CFTR ΔF508 CFBE41o– mutant polarized airway epithelial cell model

The group found that different models were successful at reproducing the biological function of P. aeruginosa, and by combining them, 96% of the reference strain PAO1 were accurately represented.

Elusive genes are poorly modeled by standard a reference strain

Despite this success, the group found that 211 genes that were unsuccessfully captured by laboratory models using PAO1 but were captured using a clinical strain. These are referred to by the team as ‘elusive genes’ and encompass genes known to change due to the accumulation of mutation.

51 of the 211 genes were captured using the CF clinical isolate grown in the SCFM2 model, which indicated that models can be improved by selecting clinical strains rather than a standard reference strain for specific functional genes.

Model accuracy depends on the functional category of genes explored

Notably, the group was surprised to find that the mouse model performance was not significantly better than other in vitro models. This is significant as mice are regarded as the gold standard for the recapitulation of the environment for human infections, presenting a good mimic of the host immune response.

The SCFM2 was a better performing model relative to the murine model, but there were distinct categories in which the murine model performed better; for example, gene encoding porins were more accurately represented in the mouse model.

Although overall accuracy scores are important, functional categories, pathways, and genes must be considered separately as better overall scoring models will not be appropriate in all cases.

Model improvements by the addition of co-occurring microbes

The framework devised also outlines strategies to improve model systems. For example, models can be modified to reflect the gene expression seen in CF sputum. In the case of SCFM2, the addition of spermidine can reduce the gene expression for polyamine biosynthesis (which is higher relative to that seen in CF sputum) to improve its accuracy.

This approach is useful for genes encoding well-categorized functions such as biosynthetic and catabolic processes. As transcriptomic data becomes available, such as gene interaction networks, genes of unknown functions can be modeled with similar accuracy.

The framework offers the potential to examine the effect of microbial modulation on P. aeruginosa gene expression. The group hypothesized that the addition of microbes may improve the accuracy of the ‘elusive genes’. Moreover, the addition of immune cell components in vitro model systems may also help to elucidate the signals that stimulate P. aeruginosa gene expression.

At present, the team only analyzed the mean accuracy of the score, but as more transcriptomes are categorized, the team hopes that the study of the distribution of these scores can be quantified.

Future direction

The team outlines the need to test strains harvested at different points during their growth phase as they represent different models, owing to their distinct expression profiles. Moreover, different labs may conduct model experiments differently and some library preparation processes have the potential to skew the accuracy of score calculations. Finally, downstream analysis can impact scores.

The team concluded by stating: “Our approach can also easily be extended in the future to community-wide functionality in polymicrobial communities, rather than simply the functions of its individual members”


Cornforth, DM et al. (2020) Quantitative Framework for Model Evaluation in Microbiology Research Using Pseudomonas aeruginosa and Cystic Fibrosis Infection as a Test Case. mBio. doi: 10.1128/mBio.03042-19

Further Reading

Last Updated: Mar 16, 2020

Hidaya Aliouche

Written by

Hidaya Aliouche

Hidaya is a science communications enthusiast who has recently graduated and is embarking on a career in the science and medical copywriting. She has a B.Sc. in Biochemistry from The University of Manchester. She is passionate about writing and is particularly interested in microbiology, immunology, and biochemistry.


Please use one of the following formats to cite this article in your essay, paper or report:

  • APA

    Aliouche, Hidaya. (2020, March 16). Devising a General Microbial Model Evaluation Framework for Microbiology Research. News-Medical. Retrieved on February 28, 2021 from

  • MLA

    Aliouche, Hidaya. "Devising a General Microbial Model Evaluation Framework for Microbiology Research". News-Medical. 28 February 2021. <>.

  • Chicago

    Aliouche, Hidaya. "Devising a General Microbial Model Evaluation Framework for Microbiology Research". News-Medical. (accessed February 28, 2021).

  • Harvard

    Aliouche, Hidaya. 2020. Devising a General Microbial Model Evaluation Framework for Microbiology Research. News-Medical, viewed 28 February 2021,


The opinions expressed here are the views of the writer and do not necessarily reflect the views and opinions of News Medical.