BACKGROUND: Set-level classification of gene expression data has received significant attention recently. In this setting, high-dimensional vectors of features corresponding to genes are converted into lower-dimensional vectors of features corresponding to biologically interpretable gene sets. The dimensionality reduction brings the promise of a decreased risk of overfitting, potentially resulting in improved accuracy of the learned classifiers. However, recent empirical research has not confirmed this expectation. Here we hypothesize that the reported unfavorable classification results in the set-level framework were due to the adoption of unsuitable gene sets defined typically on the basis of the Gene ontology and the KEGG database of metabolic networks. We explore an alternative approach to defining gene sets, based on regulatory interactions, which we expect to collect genes with more correlated expression. We hypothesize that such more correlated gene sets will enable to learn more accurate classifiers. METHODS: We define two families of gene sets using information on regulatory interactions, and evaluate them on phenotype-classification tasks using public prokaryotic gene expression data sets. From each of the two gene-set families, we first select the best-performing subtype. The two selected subtypes are then evaluated on independent (testing) data sets against state-of-the-art gene sets and against the conventional gene-level approach. RESULTS: The novel gene sets are indeed more correlated than the conventional ones, and lead to significantly more accurate classifiers. The novel gene sets are indeed more correlated than the conventional ones, and lead to significantly more accurate classifiers. CONCLUSION: Novel gene sets defined on the basis of regulatory interactions improve set-level classification of gene expression data. The experimental scripts and other material needed to reproduce the experiments are available at http://ida.felk.cvut.cz/novelgenesets.tar.gz.

Faculty of Electrical Engineering Czech Technical University Technická 2 Prague 166 27 Czech Republic

School of Computer Science and Informatics Cardiff University Queen's Buildings 5 The Parade Roath Cardiff CF24 3AA UK

Zobrazit více v PubMed

Mramor M, Toplak M, Leban G, Curk T, Zupan B. On utility of gene set signatures in gene expression-based cancer class prediction. J Mach Learn Res - Proc Track. 2010;8:55–64.

Holec M, Kléma J, Zelezný F, Tolar J. Comparative evaluation of set-level techniques in predictive classification of gene expression samples. BMC Bioinformatics. 2012;13 Suppl 1(Suppl 10):15. doi: 10.1186/1471-2105-13-S10-S15. PubMed DOI PMC

Abraham G, Kowalczyk A, Loi S, Haviv I, Zobel J. Prediction of breast cancer prognosis using gene set statistics provides signature stability and biological context. BMC Bioinformatics. 2010;11:277. doi: 10.1186/1471-2105-11-277. PubMed DOI PMC

Krejnik M, Klema J. Empirical evidence of the applicability of functional clustering through gene expression classification. IEEE/ACM Trans Comput Biol Bioinform. 2012;9(3):788–98. doi: 10.1109/TCBB.2012.23. PubMed DOI

Staiger C, Cadot S, Kooter R, Dittrich M, Müller T, Klau GW, et al. A critical evaluation of network and pathway-based classifiers for outcome prediction in breast cancer. PloS One. 2012;7(4):34796. doi: 10.1371/journal.pone.0034796. PubMed DOI PMC

Staiger C, Cadot S, Györffy B, Wessels LFA, Klau GW. Current composite-feature classification methods do not outperform simple single-genes classifiers in breast cancer prognosis. Front Genet. 2013; 4. doi:10.3389/fgene.2013.00289. PubMed DOI PMC

Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000;25(1):25–9. doi: 10.1038/75556. PubMed DOI PMC

Kanehisa M, Goto S, Kawashima S, Okuno Y, Hattori M. The KEGG resource for deciphering the genome. Nucleic Acids Res. 2004;32(Database issue):277–80. doi: 10.1093/nar/gkh063. PubMed DOI PMC

Huang DWW, Sherman BTT, Lempicki RAA. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 2008;:1–13. doi:10.1093/nar/gkn923. PubMed DOI PMC

Mitra S, Ghosh S. Feature selection and clustering of gene expression profiles using biological knowledge. IEEE Trans Syst Man Cybern Syst Hum, Part C. 2012;42(6):1590–9. doi: 10.1109/TSMCC.2012.2209416. DOI

Klema J, Soulet A, Cremilleux B, Blachon S, Gandrillon O. CBMS 2006: 19th IEEE International Symposium on Computer-Based Medical Systems. Washington, DC, USA: IEEE; 2006. Mining Plausible Patterns from Genomic Data.

Leyritz J, Schicklin S, Blachon S, Keime C, Robardet C, Boulicaut J-F, et al. SQUAT: a web tool to mine human, murine, and avian SAGE data. BMC Bioinformatics. 2008; 9(378). doi:10.1186/1471-2105-9-378. PubMed DOI PMC

Andel M, Klema J, Krejcik Z. Network-Constrained Forest for Regularized Classification of Omics Data. Methods. 2015;83:88–97. doi: 10.1016/j.ymeth.2015.04.006. PubMed DOI

Libalova H, Krckova S, Uhlirova K, Milcova A, Schmuczerova J, Ciganek M, et al. Genotoxicity but not the AhR-mediated activity of PAHs is inhibited by other components of complex mixtures of ambient air pollutants. Toxicol Lett. 2014;225(3):350–7. doi: 10.1016/j.toxlet.2014.01.028. PubMed DOI

Dostalova Merkerova M, Krejcik Z, Belickova M, Hrustincova A, Klema J, Stara E, et al. Genome-wide miRNA profiling in myelodysplastic syndrome with del(5q) treated with lenalidomide. Eur J Haematol. 2015;95(1):35–43. doi: 10.1111/ejh.12458. PubMed DOI

Xiao G, Martinez-Vaz B, Pan W, Khodursky AB. Operon information improves gene expression estimation for cDNA microarrays. BMC Genomics. 2006;7:87. doi: 10.1186/1471-2164-7-87. PubMed DOI PMC

Tintle NL, Sitarik A, Boerema B, Young K, Best AA, Dejongh M. Evaluating the consistency of gene sets used in the analysis of bacterial gene expression data. BMC Bioinformatics. 2012;13(1):193. doi: 10.1186/1471-2105-13-193. PubMed DOI PMC

Maas WK. Studies on the mechanism of repression of arginine biosynthesis in Escherichia coli. II. Dominance of repressibility in diploids. J Mol Biol. 1964;8:365–70. doi: 10.1016/S0022-2836(64)80200-X. PubMed DOI

Gutiérrez-Ríos RM, Rosenblueth DA, Loza JA, Huerta AM, Glasner JD, Blattner FR, et al. Regulatory network of Escherichia coli: consistency between literature knowledge and microarray profiles. Genome Res. 2003;13(11):2435–43. doi: 10.1101/gr.1387003. PubMed DOI PMC

Salgado H, Peralta-Gil M, Gama-Castro S, Santos-Zavaleta A, Muñiz-Rascado L, García-Sotelo JS, et al. RegulonDB v8.0: omics data sets, evolutionary conservation, regulatory phrases, cross-validated gold standards and more. Nucleic Acids Res. 2013;41(Database issue):203–13. doi: 10.1093/nar/gks1201. PubMed DOI PMC

Cortes C, Vapnik V. Support-vector networks. Mach Learn. 1995;20(3):273–97.

Perez-Rueda E. The repertoire of DNA-binding transcriptional regulators in Escherichia coli K-12. Nucleic Acids Res. 2000;28(8):1838–47. doi: 10.1093/nar/28.8.1838. PubMed DOI PMC

Edgar R, Domrachev M, Lash AE. Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. 2002;30(1):207–10. doi: 10.1093/nar/30.1.207. PubMed DOI PMC

Barrett T, Wilhite SE, Ledoux P, Evangelista C, Kim IF, et al. NCBI GEO: archive for functional genomics data sets–update. Nucleic Acids Res. 2013;41(Database issue):991–5. doi: 10.1093/nar/gks1193. PubMed DOI PMC

Keene JD, Tenenbaum SA. Eukaryotic mRNPs May Represent Posttranscriptional Operons. Molecular Cell. 2002;9(6):25–9. doi: 10.1016/S1097-2765(02)00559-2. PubMed DOI

Demšar J. Statistical Comparisons of Classifiers over Multiple Data Sets. J Mach Learn Res. 2006;7:1–30.