Information Technology Reference
In-Depth Information
Curved
Rigid
Flexible
Figure 5. Output examples of the bend.it server. A. profile plots of bendability
(blue) and curvature (red) along the 350bp L. tarantolae kinetoplast sequence.
Profile plots provide a visual aid to locate “interesting” regions along a DNA
sequence. B. correlation (2D) plot of curvature vs. bendability of the same sequence.
2. Prediction of DNA properties other than curvature
From the computational point of view curvature is a local property of DNA that can be
represented by numeric values assigned to each position of a DNA sequence. The same
philosophy can be extended to a large number of other DNA properties that can be assigned
to a short segment of DNA. There are a few common approximations underlying many
parametric descriptions: a) The property is local, i.e. a given n-mer in DNA will have the
same property irrespective of its sequence environment ("context"). This may be true for
molecular properties depending only on the nucleobases, but is a very rough approximation
for complex, statistically derived properties like conformational preferences since, for
instance, even dinucleotides are known to adopt a few different conformations depending
on their neighbours. b) Segments within DNA (nucleotides, dinucleotides) contribute
independently to a given property. This makes it possible to use simple linear or log/linear
models to experimental data.
As an example, bending propensity parameters for trinucleotides were deduced from
DNAseI digestibility vs. sequence data based on the following principles [1]. (i) Locality:
DNase I interacts with the window of 6 nucleotides around the cleaved bond and its cutting
efficiency depends only on this window. (ii) This window is represented as four
Search WWH ::




Custom Search