Cancer classification from time series microarray data through regulatory Dynamic Bayesian Networks. (January 2020)

• Record Type: Journal Article
• Authors: Kourou, Konstantina; Rigas, George; Papaloukas, Costas; Mitsis, Michalis; Fotiadis, Dimitrios I.
• Is Part Of: Computers in biology and medicine. Volume 116(2020)
• Publication Date: 2020-01
• Subjects: Genomic profiling -- Microarray data -- Dynamic Bayesian Networks -- Cancer classification -- Regulatory genes; Medicine -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; 610.285
• DOI: 10.1016/j.compbiomed.2019.103577
• Languages: English
• ISSNs: 0010-4825

Abstract

Abstract: Genomic profiling of cancer studies has generated comprehensive gene expression patterns for diverse phenotypes. Computational methods which employ transcriptomics datasets have been proposed to model gene expression data. Dynamic Bayesian Networks (DBNs) have been used for modeling time series datasets and for the inference of regulatory networks. Furthermore, cancer classification through DBN-based approaches could reveal the importance of exploiting knowledge from statistically significant genes and key regulatory molecules. Although microarray datasets have been employed extensively by several classification methods for decision making, the use of new knowledge from the pathway level has not been addressed adequately in the literature in terms of DBNs for cancer classification. In the present study, we identify the genes that act as regulators and mediate the activity of transcription factors that have been found in all promoters of our differentially expressed gene sets. These features serve as potential priors for distinguishing tumor from normal samples using a DBN-based classification approach. We employed three microarray datasets from the Gene Expression Omnibus (GEO) public functional repository and performed differential expression analysis. Promoter and pathway analysis of the identified genes revealed the key regulators which influence the transcription mechanisms of these genes. We applied the DBN algorithm on selected genes and identified the features that can accurately classify the samples into tumors and controls. Both accuracy and Area Under the Curve (AUC) were high for the gene sets comprising of the differentially expressed genes along with their master regulators (accuracy: 70.8%–98.5%; AUC: 0.562–0.985). Highlights: Cancer classification through Dynamic Bayesian Network models. Exploitation of time series microarray data. Promoter and pathway analysis of the identified differentially expressed genes. Obtained accuracy of three microarray datasets equals to 98.5%, 73.3% and 70.8%, respectively.