ABSTRACT
This chapter presents an application of genetic algorithms to a problem in molecular biology. Many proteins occurring in cells participate in biochemical events such as degradation, chemical modification, directional transport, etc. It has been shown that in certain cases, a string of amino acids serves as a specific signal; thus proteins which carry this sequence within their primary structures participate in some molecular event, while proteins lacking this sequence do not (the endoplasmic reticulum retention signal “KDEL” is a good example). Finding the sequence of a specific possible signal based only on the primary structures of a group of proteins thought to carry it is a very difficult task. No good algorithm currently exists for locating brand new signals. A genetic algorithm is described here which is able to discover such sequences. This algorithm is able to search the enormous state space of all possible signals in reasonable time, and locate likely signal sequences (which can then be tested empirically). The algorithm can also be used to find signature sequences in related proteins. Because genetic algorithms are domain independent, a parametrization study is also presented, which shows optimal values of certain constants for this specific task.54
