DNA unzipping via stopped birth and death processes with unknown transition probabilities

TL;DR

This paper introduces a probabilistic model for DNA unzipping, providing explicit formulas for prediction accuracy and extending previous models to address open questions in DNA sequencing via mechanical stimulation.

Contribution

It offers a new probabilistic framework for DNA base sequence prediction using stopped birth and death processes with unknown transition probabilities, including explicit error probability formulas.

Findings

Derived explicit error probability formulas for sequence prediction

Compared discrete and continuous time models for DNA unzipping

Extended initial models to answer physicists' open questions

Abstract

In this paper we provide an alternative approach to the works of the physicists S. Cocco and R. Monasson about a model of DNA molecules. The aim is to predict the sequence of bases by mechanical stimulations. The model described by the physicists is a stopped birth and death process with unknown transition probabilities. We consider two models, a discrete in time and a continuous in time, as general as possible. We show that explicit formula can be obtained for the probability to be wrong for a given estimator, and apply it to evaluate the quality of the prediction. Also we add some generalizations comparing to the initial model allowing us to answer some questions asked by the physicists.