High-throughput microarray technology in diagnostics of enterobacteria based on genome-wide probe selection and regression analysis
1 University of Würzburg, Institute for Molecular Infection Biology, Josef-Schneider-Str. 2/Bau D15, 97080 Würzburg, Germany
2 Department of Bioinformatics, University of Würzburg, Am Hubland, 97074 Würzburg, Germany
3 Bioinformatics for High-Throughput Technologies, Computer Science 11, TU Dortmund, 44221 Dortmund, Germany
4 Scienion AG, Volmerstraße 7b, 12489 Berlin, Germany
5 Robert Koch-Institute, Wernigerode Branch, Burgstrasse 37, 38855 Wernigerode, Germany
6 Department of Molecular Microbiology and Biotechnology, Tel Aviv University, Tel Aviv 69978, Israel
7 Faculty of Medicine Carl Gustav Carus, Institute for Medical Microbiology and Hygiene, Technology University (TU) Dresden, Dresden, Germany
8 German Academy of Sciences Leopoldina, Emil-Abderhalden-Str. 37, 06108 Halle/Saale, Germany
9 Institute for Hygiene, University of Münster, Robert-Koch-Str. 41, 48149 Münster, Germany
BMC Genomics 2010, 11:591 doi:10.1186/1471-2164-11-591Published: 21 October 2010
The Enterobacteriaceae comprise a large number of clinically relevant species with several individual subspecies. Overlapping virulence-associated gene pools and the high overall genome plasticity often interferes with correct enterobacterial strain typing and risk assessment. Array technology offers a fast, reproducible and standardisable means for bacterial typing and thus provides many advantages for bacterial diagnostics, risk assessment and surveillance. The development of highly discriminative broad-range microbial diagnostic microarrays remains a challenge, because of marked genome plasticity of many bacterial pathogens.
We developed a DNA microarray for strain typing and detection of major antimicrobial resistance genes of clinically relevant enterobacteria. For this purpose, we applied a global genome-wide probe selection strategy on 32 available complete enterobacterial genomes combined with a regression model for pathogen classification. The discriminative power of the probe set was further tested in silico on 15 additional complete enterobacterial genome sequences. DNA microarrays based on the selected probes were used to type 92 clinical enterobacterial isolates. Phenotypic tests confirmed the array-based typing results and corroborate that the selected probes allowed correct typing and prediction of major antibiotic resistances of clinically relevant Enterobacteriaceae, including the subspecies level, e.g. the reliable distinction of different E. coli pathotypes.
Our results demonstrate that the global probe selection approach based on longest common factor statistics as well as the design of a DNA microarray with a restricted set of discriminative probes enables robust discrimination of different enterobacterial variants and represents a proof of concept that can be adopted for diagnostics of a wide range of microbial pathogens. Our approach circumvents misclassifications arising from the application of virulence markers, which are highly affected by horizontal gene transfer. Moreover, a broad range of pathogens have been covered by an efficient probe set size enabling the design of high-throughput diagnostics.