4

I would like to perform wavelet smoothing at different scales in R. I got the idea from this figure (panel A) were they measure the density/intensity of a particular signal at different scales.

Here the image shows the linear coordinates along one chromosome on the x axis and the y axis shows the 'bandwidth' or bin size in which the signal was measured.

Does using wavelets here makes sense?

The first figure comes from de Wit et al. Global Chromatin Domain Organization of the Drosophila Genome, PLoS Genetics 2008.

Taken from "Global Chromatin Domain Organization of the Genome"

I found this kind of multi-scale wavelet plot but I am not sure if it is reflecting the same thing as the first figure.

This figures comes from this website

This image shoes data from tree rings

pedrosaurio
  • 1,283
  • 2
  • 14
  • 19

1 Answers1

1

I don't think that you can compare using wavelets to the 1st example. In the original thesis, it (domainogram) is described as a type of heat map with color levels proportional to probabilities of local enrichment levels of a chromatin component related to the width of neighbours sampled. A wavelet spectrogram does not measure probabilities so much as break down the energy spectrum of a composite time domain signal into both local time and frequency based bins. In the wavelet case, the levels of the heat map are related to the strength of the td signal in localised space and time.

I didn't see anything to suggest that the underlying data could be represented as a time domain signal, nor that the underlying probabilities had any relationship to a time based signal. In either case, if you are able to generate the underlying levels in a tabular format, a simple heat map in R can be used to display the data.

The authors of the paper (sourced below) stated that a copy of the original algorithm of the domainogram (written in Pearl) can be requested.

dare.uva.nl/document/188925 Ch 2.

g g
  • 26
  • 2