Whereas genomes can be rapidly sequenced, the functions of many genes are incompletely or erroneously annotated because of a lack of experimental evidence or prior functional knowledge in sequence databases. To address this weakness, we describe here a model-enabled gene search (MEGS) approach that (i) identifies metabolic functions either missing from an organism's genome annotation or incorrectly assigned to an ORF by using discrepancies between metabolic model predictions and experimental culturing data; (ii) designs functional selection experiments for these specific metabolic functions; and (iii) selects a candidate gene(s) responsible for these functions from a genomic library and directly interrogates this gene's function experimentally. To discover gene functions, MEGS uses genomic functional selections instead of relying on correlations across large experimental datasets or sequence similarity as do other approaches. When applied to the bioluminescent marine bacterium Vibrio fischeri, MEGS successfully identified five genes that are responsible for four metabolic and transport reactions whose absence from a draft metabolic model of V. fischeri caused inaccurate modeling of high-throughput experimental data. This work demonstrates that MEGS provides a rapid and efficient integrated computational and experimental approach for annotating metabolic genes, including those that have previously been uncharacterized or misannotated.

作者：Shu, Pan；Kiel, Nikolakakis；Paul A, Adamczyk；Min, Pan；Edward G, Ruby；Jennifer L, Reed

来源：The Journal of biological chemistry 2017 年 292卷 24期