Triclustering on temporary microarray data using the TriGen algorithm

Abstract

and riquelme@us.es is important, for example, cell cycles, development at the molecular level or evolution of diseases [8]. Therefore is necessary to develop specific tools for data analysis in which genes are evaluated under certain conditions considering the time factor. In this context we present the TriGen algorithm, which goes a step further than clustering and biclustering techniques in the creation of groups of pattern similarity for genes. TriGen works on a three-dimensional space, thus taking into account the time factor, and allowing the evaluation of the behavior of genes only under certain conditions and only under certain time points. TriGen applies an evolutionary technique, genetic algorithms, to find solutions that we refer to as triclusters. Other works related with this approach are in [9] and [10]. The rest of the paper is structured as follows: Section II describes the algorithm in detail, Section III shows the results using both synthetic and real data. Section IV summarizes the conclusions reached and proposals for future work. II. METHODOLOGYWe describe the implementation of the TriGen algorithm. In this section we explain the inputs and outputs of the algorithm and we provide a detailed description of the evolutionary process and all the operators implied. A. Input dataThe input data is obtained from temporal microarray experiments. Each of these microarrays reveals the expression level under specific experimental conditions and at an instant of time. Therefore, the input data consists of T number of microarrays, as many as time points to be analyzed. Each value of a microarray for an specific time t represents the level of gene expression of a gene g under a specific experimental condition c. B. Definition of TriclusterWe define a tricluster as a subset of time points T , a subset of genes G and a subset of conditions C extracted from the input data. In this particular work, each tricluster contains the expression values of the these three sets and a fitness value that indicates the tricluster's quality. The fitness function will be described in detail in Section II-C6. Qualitatively, a tricluster will provide information on behavior pattern of a 877Abstract-The analysis of microarray data is a computational challenge due to the characteristics of these data. Clustering techniques are widely applied to create groups of genes that exhibit a similar behavior under the conditions tested. Biclustering emerges as an improvement of classical clustering since it relaxes the constraints for grouping allowing genes to be evaluated only under a subset of the conditions and not under all of them. However, this technique is not appropriate for the analysis of temporal microarray data in which the genes are evaluated under certain conditions at several time points. In this paper, we propose the TriGen algorithm, which finds triclusters that take into account the experimental conditions and the time points, using evolutionary computation, in particular genetic algorithms, enabling the evaluation of th...