Collection Of Biological Sequences Research Articles

A cost-aggregating integer linear program for motif finding

In the motif finding problem one seeks a set of mutually similar substrings within a collection of biological sequences. This is an important and widely-studied problem, as such shared motifs in DNA often correspond to regulatory elements. We study a combinatorial framework where the goal is to find substrings of a given length such that the sum of their pairwise distances is minimized. We describe a novel integer linear program for the problem, which uses the fact that distances between substrings come from a limited set of possibilities allowing for aggregate consideration of sequence position pairs with the same distances. We show how to tighten its linear programming relaxation by adding an exponential set of constraints and give an efficient separation algorithm that can find violated constraints, thereby showing that the tightened linear program can still be solved in polynomial time. We apply our approach to find optimal solutions for the motif finding problem and show that it is effective in practice in uncovering known transcription factor binding sites.

Self-Organizing Maps of Position Weight Matrices for Motif Discovery in Biological Sequences

The identification of overrepresented motifs in a collection of biological sequences continues to be a relevant and challenging problem in computational biology. Currently popular methods of motif discovery are based on statistical learning theory. In this paper, a machine-learning approach to the motif discovery problem is explored. The approach is based on a Self-Organizing Map (SOM) where the output layer neuron weight vectors are replaced by position weight matrices. This approach can be used to characterise features present in a set of sequences, and thus can be used as an aid in overrepresented motif discovery. The SOM approach to motif discovery is demonstrated using biological sequence datasets, both real and simulated