Approaching miRNA family classification through constructive neural networks
ARTIGO
Inglês
Agradecimentos: The authors thank FAPESP and CNPq for their support
Abstract: MicroRNAs (or miRNAs) are non-coding RNA molecules associated to gene expression, acting in post-transcriptional regulation, either by inhibiting the translation of messenger RNAs or by promoting its degradation. Since the discovery of the first miRNA, a considerable number of miRNAs,...
Abstract: MicroRNAs (or miRNAs) are non-coding RNA molecules associated to gene expression, acting in post-transcriptional regulation, either by inhibiting the translation of messenger RNAs or by promoting its degradation. Since the discovery of the first miRNA, a considerable number of miRNAs, found in plants, worms, vertebrates, etc. have been described in the literature. The experimental determination of their corresponding functionality, however, has been restricted to a much smaller number. A considerable volume of miRNA sequences have been categorized into families, based on their associated mature sequence or on the structure of their pre-MicroRNA. Considering that members of the same family have the tendency of biologically function in similar ways, a structured family can help to detect functions still not associated with existing families or, via an induced classifier, assign potential candidates to families, based on their feature values. This paper investigates the use of five constructive neural network (CoNN) algorithms, the Tower, Pyramid, Tiling, Perceptron Cascade and the BabCoNN, in the miRNA knowledge domain, aiming at evaluating how well they perform on MicroRNA data, as far as classification of new miRNA sequences is concerned. Also, the paper comparatively discusses the CoNN results with those obtained using the Multi-layer Perceptron (MLP), Random Forest (RF) and Support Vector Machines (SVM) algorithms. Results show that CoNN algorithms are an efficient alternative for miRNA classification when a suitable combination of dimensionality reduction algorithm and number of dimensions for describing the data is considered
CONSELHO NACIONAL DE DESENVOLVIMENTO CIENTÍFICO E TECNOLÓGICO - CNPQ
FUNDAÇÃO DE AMPARO À PESQUISA DO ESTADO DE SÃO PAULO - FAPESP
Fechado
DOI: https://doi.org/10.1109/IJCNN.2018.8489019
Texto completo: https://ieeexplore.ieee.org/document/8489019
Approaching miRNA family classification through constructive neural networks
Approaching miRNA family classification through constructive neural networks
Fontes
Proceedings of the 2018 International Joint Conference on Neural Networks Piscataway, NJ : Institute of Electrical and Electronics Engineers, 2018. p. 4280-4287 |