fractal and entropy based sequence visualization

[krawitz]

Hi folks,

in the early days of sequencing there were several fractal and entropy based visualization algorithms. These tools were very handy to get a quick grasp for patterns in the DNA sequences.
I would like to analyze some of my reads (especially those, that don't map) with such algorithms.
Does anyone know if there are already descent implementations of such algorithms for short read NGS data?

Cheers,

Peter

[Xi Wang]

I think you need motif discovery algorithms to get the consensus DNA sequence, or a weight matrix. But you also should to know, those algorithms are very time consuming, especially when you have a huge amount of short reads.

Why you need such an application? For ChIP-seq data?

[krawitz]

No I don't need it for ChIP Seq data. I just want to analyse some of my reads that don't map against any reasonable reference sequence. The reads that I am talking about look somehow repetitive, or of low sequence complexity, but they are definitely not from repeat regions.
Thus I wanted to analyse them with algorithms similar to the ones e.g. described in "Dynamical Visualization of the DNA sequence and its nucleotide content" by Pasechnik et al.

[Xi Wang]

I see. It is a nice tool. Did you get it from the authors?

[Wolfgang Huber]

The HilbertVis tool by Simon Anders might be useful - it allows visualisation of a long vector through a space filling curve motivated by Hilbert or Peano curves: S. Anders: “Visualisation of genomic data with the Hilbert curve”, Bioinformatics, Vol. 25 (2009) pp. 1231-1235

[Wolfgang Huber]

Peter and Xi,

on closer reading of your posts, I realised that you are looking for something different. Sorry for the noise. Perhaps my post is still useful for something or -body.

Btw, making a plot like Figs. 3 and 4 in the Pasechnik et al. paper should be quite straightforward with R using the ShortRead and Biostrings package.