Model selection and sensitivity analysis for sequence pattern models

TL;DR

This paper introduces a MAP-based approach for selecting models in motif discovery, analyzing its asymptotic correctness and robustness to prior choices, with guidelines for hyper-parameter tuning.

Contribution

It proposes a novel MAP criterion for model selection in motif discovery and studies its asymptotic properties and robustness to prior hyper-parameters.

Findings

MAP asymptotically predicts correct model size

Guidelines for choosing prior hyper-parameters

Analysis of robustness to prior specification

Abstract

In this article we propose a maximal a posteriori (MAP) criterion for model selection in the motif discovery problem and investigate conditions under which the MAP asymptotically gives a correct prediction of model size. We also investigate robustness of the MAP to prior specification and provide guidelines for choosing prior hyper-parameters for motif models based on sensitivity considerations.