A new method for detecting signal regions in ordered sequences of real numbers, and application to viral genomic data

PLoS One. 2018 Apr 13;13(4):e0195763. doi: 10.1371/journal.pone.0195763. eCollection 2018.

Abstract

We present a fast, robust and parsimonious approach to detecting signals in an ordered sequence of numbers. Our motivation is in seeking a suitable method to take a sequence of scores corresponding to properties of positions in virus genomes, and find outlying regions of low scores. Suitable statistical methods without using complex models or making many assumptions are surprisingly lacking. We resolve this by developing a method that detects regions of low score within sequences of real numbers. The method makes no assumptions a priori about the length of such a region; it gives the explicit location of the region and scores it statistically. It does not use detailed mechanistic models so the method is fast and will be useful in a wide range of applications. We present our approach in detail, and test it on simulated sequences. We show that it is robust to a wide range of signal morphologies, and that it is able to capture multiple signals in the same sequence. Finally we apply it to viral genomic data to identify regions of evolutionary conservation within influenza and rotavirus.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Algorithms
  • Biological Evolution
  • Computational Biology / methods*
  • Computer Simulation
  • Databases, Genetic
  • Genetic Variation
  • Genome, Viral*
  • Genomics / methods*
  • Humans

Grants and funding

JPS is funded by a National Institute for Health Research Academic Clinical Fellowship. This article/paper/report presents independent research funded by the National Institute for Health Research (NIHR). The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health. Work in the laboratory of AMLL is supported by the Biomedical Research Centre.