Analysing the localisation sites of proteins through neural networks ensembles

摘要: Scientists involved in the area of proteomics are currently seeking integrated, customised and validated research solutions to better expedite their work analyses drug discoveries. Some drugs most cell targets proteins, because proteins dictate biological phenotype. In this context, automated analysis protein localisation is more complex than DNA sequences; nevertheless benefits be derived same or greater importance. order accomplish target, right choice kind methods for these applications, especially when data set drastically imbalanced, very important crucial. paper we investigate performance some commonly used classifiers, such as K nearest neighbours feed-forward neural networks with without cross-validation, a class imbalanced problems from bioinformatics domain. Furthermore, construct ensemble-based schemes using notion diversity, empirically test on problems. The experimental results favour generation network ensembles able produce good generalisation ability significant improvement compared other single classifier methods.