Evoscan : a probabilistic framework for gene annotation using comparative genomics

Abstract

Evoscan is based on a (pseudo higher-order) evolutionary hidden Markov model that integrates phylogenetic information into 'Genscan" code based on Twinscan. Its three major components consist of probabilistic models of both the DNA sequence, phylogenetic information of two or more closely related organisms as well as a dynamic programming framework. Evoscan software has an object oriented design and is implemented using ANSI C/C++ and makes use of public domain source code of Twinscan and Evogene softwares. In this implementation, the initial results indicate to improve the prediction accuracy.