IDENTIFICATION OF RIBOSOME BINDING-SITES IN ESCHERICHIA-COLI USING NEURAL-NETWORK MODELS

This study investigated the use of neural networks in the identificati on of Escherichia coil ribosome binding sites. The recognition of thes e sites based on primary sequence data is difficult due to the multipl e determinants that define them. Additionally, secondary structure pla ys a significant role in the determination of the site and this inform ation is difficult to include in the models. Efforts to solve this pro blem have so far yielded poor results. A new compilation of E.coli rib osome binding sites was generated for this study. Feedforward backprop agation networks were applied to their identification. Perceptrons wer e also applied, since they have been the previous best method since 19 82. Evaluation of performance for all the neural networks and perceptr ons was determined by ROC analysis. The neural network provided signif icant improvement in the recognition of these sites when compared with the previous best method, finding less than half the number of false positives when both models were adjusted to find an equal number of ac tual sites. The best neural network used an input window of 101 nucleo tides and a single hidden layer of 9 units. Both the neural network an d the perceptron trained on the new compilation performed better than the original perceptron published by Stormo et al. in 1982.