Recognizing patterns in protein sequences using iteration-performing calculations in genetic programming
John R. Koza · 2002
Uses genetic programming with automatically defined functions (ADFs) for the dynamic creation of a pattern-recognizing computer program consisting of initially-unknown detectors, an initially-unknown iterative calculation incorporating the as-yet-undiscovered detectors, and an initially-unspecified final calculation incorporating the results of the as-yet-unspecified iteration. The program's goal is to recognize a given protein segment as being a transmembrane domain or non-transmembrane area of the protein. Genetic programming with automatic function definition is given a training set of differently-sized mouse protein segments and their correct classification. Correlation is used as the fitness measure. Automatic function definition enables genetic programming to dynamically create subroutines (detectors). A restricted form of iteration is introduced to enable genetic programming to perform calculations on the values returned by the detectors. When cross-validated, the best genetically-evolved recognizer for transmembrane domains achieves an out-of-sample correlation of 0.968 and an out-of-sample error rate of 1.6%. This error rate is better than that recently reported for five other methods.>