Title :
Signal peptide prediction on DNA sequences with artificial neural networks
Author :
Hatzigeorgiou, A.G. ; Reckzo, M.
Author_Institution :
Dept. of Genetics, Pennsylvania Univ., Philadelphia, PA, USA
Abstract :
In this work we apply the signal peptide prediction Sigfind on DNA sequences. A combination with the DIANA-TIS method for translation initiation start site (TIS) prediction can identify half of the ambiguous cases with alternative start sites correctly. Sigfind is based on the bidirectional recurrent neural network (BRNN) architecture. DIANA-TIS makes use of two modules, one sensitive to the conserved motif and one sensitive to the coding/non-coding potential around the start codon. Both modules are based on hierarchical artificial neural networks.
Keywords :
DNA; medical signal processing; molecular biophysics; prediction theory; proteins; recurrent neural nets; DIANA-TIS method; DNA sequences; artificial neural networks; bidirectional recurrent neural network; coding potential; hierarchical artificial neural networks; noncoding potential; signal peptide prediction Sigfind; translation initiation start site; Artificial neural networks; Bioinformatics; DNA; Genomics; Humans; Peptides; Proteins; Recurrent neural networks; Sequences; Signal processing;
Conference_Titel :
Biomedical Circuits and Systems, 2004 IEEE International Workshop on
Print_ISBN :
0-7803-8665-5
DOI :
10.1109/BIOCAS.2004.1454178
