SSP'05 -- Proceedings

	IEEE/SP 13th workshop on Statistical Signal Processing July, 17-20, 2005 - Bordeaux - France

Information regarding the paper

Title

Feature Extraction for DNA Base-Calling Using NNLS

Author(s)

Lucio Andrade-Cetto	Northeastern University
Elias Manolakos	Northeastern University

Get the paper in PDF format

To obtain Acrobat Reader (version 5 minimum required) necessary to his read.

Abstract

In this paper we present the first features extraction stage of a novel statistical base-caller grounded on principles of probabilistic graph theory. The feature extraction stage tries to identify landmarks in the raw traces that represent true DNA bases. The proposed peak segmentation approach addresses effectively the problem of merged peaks which is very common towards the end of the chromatogram due to the imminent loss of electrophoretic resolution. We introduce an algorithm based on non-negative least squares (NNLS)for unmixing kernels of representative peaks. To improve the robustness of the algorithm we first estimate the expected distance between bases and the expected diffusion of peaks. After testing our pre-processing, feature extraction, and base-calling approach we have found that it can achieve 25% less errors than the popular Phred base-caller for long read-length sequences (>800bp) using a large size pool of M13mp18 chromatograms.