Three factors underlying incorrect in silico predictions of essential metabolic genes

Abstract

Background: The indispensability of certain genes in an organism is important for studies of microorganism physiology, antibiotic targeting, and the engineering of minimal genomes. Time and resource intensive genome-wide experimental screens can be conducted to determine which genes are likely essential. For metabolic genes, a reconstructed metabolic network can be used to predict which genes are likely essential. The success rate of these predictions is less than desirable, especially with regard to comprehensively locating essential genes.

Results: We show that genes that are falsely predicted to be non-essential (for growth) share three characteristics across multiple organisms and growth media. First, these genes are on average connected to fewer reactions in the network than correctly predicted essential genes, suggesting incomplete knowledge of the functions of these genes. Second, they are more likely to be blocked (their associated reactions are prohibited from carrying flux in the given condition) than other genes, implying incomplete knowledge of metabolism surrounding these genes. Third, they are connected to less overcoupled metabolites.

Conclusion: The results presented herein indicate genes that cannot be correctly predicted as essential have commonalities in different organisms. These elucidated failure modes can be used to better understand the biology of individual organisms and to improve future predictions.

Figure 1

Essential genes in E. coli. This pie chart classifies all genes based on their essentiality on glucose minimal medium. The genes that are actually essential are in the green and orange regions, indicating correct and incorrect predictions, respectively. The genes that are not essential are in the blue and red regions, again indicating correct and incorrect predictions, respectively. The definitions for TP, FN, TN, FP are shown in a color-coded table for clarity.

Figure 2

The mean connectivity of genes for each organism and media condition, plus the mean connectivity for TP and FN genes separately. On average, FN genes are less connected than TP genes, meaning that they are responsible for catalyzing fewer reactions.

Figure 3

The fraction of FN and non-FN genes blocked and completely blocked. In most cases here, FN genes are more likely to be blocked (at least one of their reactions cannot be used) and completely blocked (none of their reactions can be used).

Figure 4

The mean overcoupling count for TP and FN genes, plus the FN count corrected for unequal connectivity of TP and FN genes. FN genes are on average connected to less overcoupled metabolites.