CONSTRAINED MARKOV NETWORKS FOR AUTOMATED-ANALYSIS OF G-BANDED CHROMOSOMES

Automated analysis of chromosome band patterns using probabilistic Mar kov networks has been reported in previous work. Band patterns are rep resented as strings of symbols. Inferred from a set of learning string s, a Markov network is a model of intraband and interband relations in these strings. The inference is entirely data-driven and is accomplis hed using dynamic programming. This paper presents a new model of chro mosome band patterns, the constrained Markov network, which is a speci al case of its predecessor. Substantial experimental evidence of the s uperiority of the new model over the old is given in terms of equal re sults in centromere finding and improved results in classification for the 22 autosomes. Furthermore, a method for simplification of constra ined Markov networks is shown to be of considerable importance with re spect to computational complexity.