Definition and implementation of a procedure to obtain up-todate non redundant user defined databases of DNA sequences for the identification of splicing site prediction models in human