Microarray analysis workflow based on a genetic algorithm to discover potential hub genes

CARBALLIDO, JESSICA ANDREA

CURRENT BIOINFORMATICS

BENTHAM SCIENCE PUBL LTD

Año: 2022 vol. 17

1574-8936

Abstract: The search for marker genes associated with different pathologies traditionally begins withsome form of differential expression analysis. This step is essential in most functional genomics´ worksthat analyze gene expression data. In the present article, we present a different analysis, starting fromthe known biological significance of different groups of genes and then assessing the proportion of dif-ferentially expressed genes. The analysis is performed in the context of cancer expression data to unveilthe true importance of differential expression, approaching it from different research objectives. Firstly,it was seen that the percentage of differentially expressed genes is generally low concerning gene setsannotated in KEGG. On the other hand, it was observed that in the training and prediction process ofboth statistical and machine learning models, the fact of using differentially expressed genes sustainablyimproves their results.