Neuroevolution as a tool for microarray gene expression pattern identification in cancer research

摘要: Microarrays are still one of the major techniques employed to study cancer biology. However, identification expression patterns from microarray datasets is a significant challenge overcome. In this work, new approach using Neuroevolution, machine learning field that combines neural networks and evolutionary computation, provides aid in by simultaneously classifying data selecting subset more relevant genes. The main algorithm, FS-NEAT, was adapted addition structural operators designed for high dimensional data. addition, rigorous filtering preprocessing protocol select quality proposed method, 13 three different types. results show Neuroevolution able successfully classify samples when compared with other methods literature, while also finding subsets genes can be generalized algorithms carry biological information. This detected 177 genes, 82 were validated as already being associated their respective types 44 cancer, becoming potential targets explored biomarkers. Five long non-coding RNAs detected, which four don't have described functions yet. found intrinsically related extracellular matrix, exosomes cell proliferation. obtained work could unraveling molecular mechanisms underlying tumoral process describe future works.