Finding Genes in DNA Using Decision Trees and Dynamic Programming

This study demonstratesthe use of decision tree classifiers as the basis for a general gene-finding system. The system uses a dynamic programmingalgorithm that. finds the optimal segmentation of a DNAsequence into coding and noncodingregions (exonsand introits). ]'he optimality property is dependentoll a separate scoring function that takes a subsequenceand assigns to it a score reflecting the probability that the sequence is an exon. In this study, the scoring functions were sets of decision trees and rules that were combinedto give the probability estimate. Experimental results on a newly collected database of humanDNAsequences are encouraging, and some newobservations about the structure of classifiers for tile gene-finding problem have emerged from this study. Wealso provide descriptions of a newprobability chain modelthat producesvery accurate filters to find donorand acceptor sites.

• Type: Conference
• Year: 1996
• Where: ISMB
• Authors: Steven Salzberg, Xin Chen, John Henderson, Kenneth H. Fasman
• Keywords: Computational Biology, Decision Tree Classifiers, Decision Trees, ISMB 1996, Scoring Functions