version 0.8.2

[Davide Cittaro]

Hi there, 

I've just uploaded dspchip 0.8.2 (http://dspchip.googlecode.com/files/dspchip-0.8.2.tar.gz). I will modify wiki pages about dspchip options ASAP, meanwhile be aware that thresholding is no more a pipeline step and it is a step to help peak finding. Threshold is "None" by default, one has to set it explicitly to "otsu" or any future implementation I will add. Thresholding is performed to isolate foreground signal from background. Peaks are then defined as regions that pass the threshold (if thresholding is not enabled, an edge detector, much like the Canny-edge detector, is engaged).

I'm working on public data to have some benchmarks suitable for a decent publication... :-) Any good idea is welcome!

d

/*

Davide Cittaro, PhD

Cogentech - Consortium for Genomic Technologies

via adamello, 16

20139 Milano

Italy

tel.: +39(02)574303007

e-mail: davide....@ifom-ieo-campus.it

*/

[Tao Liu]

Hi Davide,

If you are looking for some broad histone marks ChIP-seq, ENCODE project has a lot of data released <http://genome.ucsc.edu/ENCODE/downloads.html>. I am not sure if they are suitable to be used in method comparison. Note that ENCODE datasets have different embargo date.

As for transcription factors, people always use motif to judge if the peaks are correct. However, for broad features, as far as I know, people do not have a clear way to compare. You may need to do some genome feature association study. For example, some features are well-known to be associated with gene activation, some are associated with gene repression, and so on.

HTH,
Tao

[Davide Cittaro]

Hi Tao,

On Feb 24, 2011, at 7:11 AM, Tao Liu wrote:

Hi Davide,

If you are looking for some broad histone marks ChIP-seq, ENCODE project has a lot of data released <http://genome.ucsc.edu/ENCODE/downloads.html>. I am not sure if they are suitable to be used in method comparison. Note that ENCODE datasets have different embargo date.

Those are among the "candidate data" we were thinking of... 

As for transcription factors, people always use motif to judge if the peaks are correct. However, for broad features, as far as I know, people do not have a clear way to compare. You may need to do some genome feature association study. For example, some features are well-known to be associated with gene activation, some are associated with gene repression, and so on.

HTH,
Tao

That helps, thanks. As you may have noticed, dspchip is not meant for TF analysis, not in the first place (although we are going to run the benchmark described in "Rye and Sætrom…. A manually curated ChIP-seq benchmark demonstrates room for improvement in current peak-finder programs. Nucleic Acids Res (2010)", just to see what happens).