Palindrome distributions and their applications

Abstract

We analyze DNA palindromes in the Coronavirus and Herpesvirus families. Specifically we study two problems. Problem 1 deals with the overall count of palindromes of a certain length in a genome where we compare the observed number of palindromes of a certain length against its expected number under Markov chain sequence models of the genome. We derive expressions for the mean and standard deviation of the number of palindromes. The resulting z-score enables us to explore whether the observed number of palindromes of a certain length is over- (or under-)represented.Problem 2 deals with a measure of local clusters of nearly palindromes at or above a certain length. This measure leads to a statistical procedure to predict the replication origins of these viruses.