Sequencing from compomers in the presence of false negative peaks

One of the main endeavors in today’s Life Science remains the efficient sequencing of long DNA molecules. Today, most de-novo sequencing of DNA is still performed using electrophoresis-based Sanger Sequencing, based on the Sanger concept of 1977. Methods using mass spectrometry to acquire the Sanger Sequencing data are limited by short sequencing lengths of 15–25 nt. Recently, we proposed a new method for DNA sequencing using base-specific cleavage and mass spectrometry, that appears to be a promising alternative to classical DNA sequencing approaches. This leads to the combinatorial problem of Sequencing From Compomers (SFC) and, finally, to the graph-theoretical problem of finding a walk in a subgraph of the de Bruijn graph. Simulations indicate that this method might be capable of sequencing DNA molecules with 200+ nt. But the way the Sequencing From Compomer Problem is formulated, it does not take into account the problem of false negative peaks that is common for real-world data: Even though an in silico simulation predicts a peak to be present in a mass spectrum, it is absent from the measured mass spectrum. We may evade this problem by choosing a very sensitive peak detection algorithm, minimizing the number of false negative peaks. Still, a single false negative peak is usually sufficient to prohibit reconstruction of the correct DNA sequence by SFC. Here, we show how to extend SFC as well as sequencing graphs to deal with false negative peaks. In addition, we present a branch-and-bound algorithm to find all sequences that agree with the sample mass spectra with the exception of a certain number of false negative peaks. Simulation results indicate that even in the presence of several false negative peaks, the presented method might be capable of sequencing DNA molecules of length 200 nt. Contact: Sebastian Boecker AG Genominformatik Technische Fakultaet Universitaet Bielefeld PF 100 131 33501 Bielefeld Germany [email protected] Date: September 12, 2003. Sebastian Böcker is currently supported by “Deutsche Forschungsgemeinschaft” (BO 1910/1-1) within the Computer Science Action Program.