Letters to the editor should be sent to:

patlacey@asmusa.org

Links to Other ASM Pages:



• Table of Contents
• ASM News Issues

A More Palatable H. pylori: Iterative Model-Building through the Peer-Review Process

High-throughput technologies have necessitated the development of mathematical models to describe the integrated function of complex biological systems (T. Ideker, T. Galitski, and L. Hood, Annu. Rev. Genomics Hum. Genet. 2:343-72, 2001). Such models differ from traditional mechanistic models in that they welcome disproof. While these comprehensive metabolic models will likely not be disproven in their entirety, they must be constantly updated and revised piecewise as new experimental information comes to light. The building of such models is therefore an iterative and a collaborative process, in which in vivo and in vitro experimental data is reconciled with the predictions of in silico computations (M. W. Covert, C. H. Schilling, I. Famili, J. S. Edwards, I. I. Goryanin, E. Selkov, and B. O. Palsson, Trends Biochem. Sci. 26:179-186, 2001, and H. Kitano, Science 295:1662-1664, 2002). Interestingly, such an iterative model building process took place through the peer-review process for the Helicobacter pylori 26695 metabolic model and derived systemic properties that appeared in a recent issue of the Journal of Bacteriology (184:4582-4593).

The in silico H. pylori model and calculations presented were the result of three stages of development, having twice benefited from the peer review process prior to publication. After the first submission, the reviewers suggested (among other things) that the de novo synthesis of adenine be included in the model based on certain physiological studies, although the genomic evidence for the pathway was lacking. A second set of reviewers later suggested that fumarate reductase identified in the genome be also treated as a succinate dehydrogenase. In each of these (and other) cases, the model was changed accordingly and the calculated predictions of the model were found to be a more accurate representation of H. pylori behavior. The peer review process therefore made a substantial unrecognized contribution to this in silico H. pylori model.

In certain cases, the reviewers' comments also raised new questions, which have yet to be explored. For example, one reviewer question inclusion of malate synthase, the second half of the glyoxylate bypass, with isocitrate lyase (the first half) asserted to be absent. Our subsequent analysis indicated that the malate synthase reaction was essential to our H. pylori network to convert the glyoxylate produced as a by-product of folate biosynthesis. This nonintuitive finding warrants further experimental investigation.

Is the resultant H. pylori model finished? Of course not; much remains to be done. We naturally hope that interested readers will attempt to use the in silico model and predictions presented, and identify areas where the model needs to be corrected or updated in light of new information, leading to a more accurate metabolic model of H. pylori. Accordingly, we wish to thank our reviewers for setting the example with their constructive, or perhaps more accurately, their reconstructive evaluations of our work, and also hope to encourage "wet" and "dry-lab" biologists to continue working together in building and curating comprehensive mathematical models of biological systems.

M. W. Covert
B. O. Palsson
University of California, San Diego

C. H. Schilling
Genomatica, Inc.
San Diego, Calif.