Information Technology Reference
In-Depth Information
Prediction and visualization of DNA
structural properties from sequence
Kristian VLAHOVICEK, László KAJÁN Sándor PONGOR
International Centre for Genetic Engineering and Biotechnology (ICGEB), Area Science
Park, Padriciano 99, 34012 Trieste, Italy
Correspondence: S. Pongor, Tel.: +39-040-3757300; Fax: +39-040-226555; E-mail:
pongor@icgeb.org
Abstract.
Visualisation of local DNA conformation is a useful tool in interpreting
and designing experiments at the molecular level. There are a number of methods
whereby local curvature as well as other conformational parameters can be
predicted. Calculation of these parameters on a genomic scale may help to clarify
the role of these elements in genomic architecture.
Introduction
Simple methods that can guide experimenters to find conspicuous regions in DNA are of
considerable interest in view of the genomic sequence generated. Structural properties,
such as flexibility or intrinsic curvature that are not consequences of explicit sequence
motifs are of particular interest since these cannot be identified from sequence similarity
searches. Over the past years our group has been developing and testing simple mechanic
models that can describe the local behaviour of DNA in such short segments, in a sequence
dependent fashion [1-9]. These methods have been extended to the calculation and
visualization of various parameters other than curvature [10], and included into WWW-
based server programs located on the ICGEB web site.
Parametric visualization of DNA characteristics consists in mapping of numerical
data to visually presentable models. The simplest form of parametric visualization is the
sequence plot i.e. a graph in which numeric values are assigned to positions along the DNA
sequence. The advantage of comparing sequence plots rather than primary sequences
originates from the simple fact that plots, unlike primary sequences, can be subjected to
arithmetic operations (averaging, subtraction, etc.) and their similarities can be
characterized in quantitative terms such as correlation coefficients and standard deviations.
This is essentially a parametric approach of sequence comparison which makes it possible,
e.g., to compare groups of sequences, to carry out a semi-quantitative comparison (ranking)
of sequences in structural terms, etc. using simple programs. The parametric visualization
of DNA sequences uses the same properties on a qualitative basis, and the conspicuous
segments can be identified by 1D, 2D or 3D plots of various parameters.
In this chapter we first describe DNA curvature as the paradigmatic concept,
followed by a short description of the server algorithms. The last part of this chapter gives
examples of applications.
