Qualitative Modeling of RNA Structure

Abstract

Determining the folding structure of an RNA molecule from its und erlying lin ea r sequence is a complex problem involving both spatial reasoning and th e use of knowledge of chemistry and biology. Most research in qualitative physics has traditionally focused on certain types of processes only, and does not provide a method for properly modeling folding. This paper introduces a new AI method for reasoning abo ut the folding process. We present a discrete model for predicting a folding structure and for qualitatively simulating the process of the s tructur e form at ion over time. We use a simple and approximate model for the task, but accuracy in the prediction is achieved by capturing knowledge from several sources. Th e modell~as been impl emented in a working program and ha s been successfully tested on several types of RNA s, including RNAs whose structures have not been fully determined yet.