Perez-Rodriguez, Javier , Arroyo-Pena, Alexis G. , Martinez-Luna, Javier , Ortuno, F , Rojas, I

Si

Proceedings Iwbbio 2013: International Work-Conference On Bioinformatics And Biomedical Engineering

Proceedings Paper

Científica

✗

✗

01/01/2013

000322416100075

Current advances in DNA sequencing technology have motivated the investigation of reliable gene identification methods, an open area currently present in bioinformatics. Gene recognition can be considered as a search problem, where many evidence sources should be combined in a scoring function that must be maximized to obtain the most likely and right gene structure in a genomic sequence. In this article, we combine a support vector machine classifier to reduce the search space together with a multiobjective genetic algorithm as main search engine to deal with a set of prospect structures. We made use of various content statistics that are commonly employed to obtain evidences of coding regions in DNA sequences which will determine the probability of a certain structure to be an actual gene. We use the human sequences located at the chromosomes 3, 19 and 21 as training set, and the chromosome 18 genes to check the performance of our system. Very promising results are obtained.