Markov Encoding for Detecting Signals in Genomic Sequences
IEEE/ACM Transactions on Computational Biology and Bioinformatics (TCBB)
Predicting Protein-Ligand Binding Site Using Support Vector Machine with Protein Properties
IEEE/ACM Transactions on Computational Biology and Bioinformatics (TCBB)
| Hi-index | 0.00 |
A modular system of neural networks is used to identify genes in DNA sequences of eukaryotic organisms. The identification task is decomposed into the detection of distinct signals using separate neural network modules. Such signals are coding regions, splice sites and transcription start regions (cap-site). A focus of the work is the use of back-percolation, cascade correlation, and time-delay neural networks. These give, in this particular application, better generalization than the well known backpropagation algorithm. This system achieves a prediction accuracy comparable to the traditionally designed gene identification packages and is able to produce more accurate protein sequences from the constructed gene structures.