Robust gene signatures from microarray data using genetic algorithms enriched with biological pathway keywords

Abstract

Genetic algorithms are widely used in the estimation of expression profiles from microarrays data. However, these techniques are unable to produce stable and robust solutions suitable to use in clinical and biomedical studies. This paper presents a novel two-stage evolutionary strategy for gene feature selection combining the genetic algorithm with biological information extracted from the KEGG database. A comparative study is carried out over public data from three different types of cancer (leukemia, lung cancer and prostate cancer). Even though the analyses only use features having KEGG information, the results demonstrate that this two-stage evolutionary strategy increased the consistency, robustness and accuracy of a blind discrimination among relapsed and healthy individuals. Therefore, this approach could facilitate the definition of gene signatures for the clinical prognosis and diagnostic of cancer diseases in a near future. Additionally, it could also be used for biological knowledge discovery about the studied disease.

Use of q-values to Improve a Genetic Algorithm to Identify Robust Gene Signatures

“…In this paper, the strategy proposed in [7] is used with some slight changes. The strategy consists of two separate stages:…”

“…A common feature of such methods is instability of results with high variability of identified features when repeated executions of the algorithm are made. To tackle this problem, recent works have proposed different methodologies that try to achieve robust feature subset selections with good performance rates in test data [7,10].Use of statistical tests with multiple features against some null hypothesis is common practice with the expectation that a proportion of such features would be incorrectly considered significant [8]. In such circumstances it is important to use some form of false discovery rate technique to either adjust the p-values [1] or use a different measure which takes into account false positives such as the q-value [8].…”

“…Use of such a measure allows focus to be placed on features which can be considered to satisfy a null hypothesis in further analysis. In the original paper [8] this methodology reduced the number of features identified in the Hedenfalk dataset from 605 to 162 within a total feature set of 3170.In this paper a modified t-test and q-values [8] are incorporated into a feature selection procedure similar to the genetic algorithm (GA) described in [7] with the purpose of identifying genes that are significant in differentiating lung cancer microarray expressions. In their approach, biological information from KEGG [5, 6] database was included into the GA resulting in more robust feature subsets with good performance…”

“…A common feature of such methods is instability of results with high variability of identified features when repeated executions of the algorithm are made. To tackle this problem, recent works have proposed different methodologies that try to achieve robust feature subset selections with good performance rates in test data [7,10].…”

“…In this paper a modified t-test and q-values [8] are incorporated into a feature selection procedure similar to the genetic algorithm (GA) described in [7] with the purpose of identifying genes that are significant in differentiating lung cancer microarray expressions. In their approach, biological information from KEGG [5,6] database was included into the GA resulting in more robust feature subsets with good performance rates.…”