Using Generalized Additive Models for bias correction in ChIP-seq data

2015 Biology of the Genomes conference at Cold Spring Harbor, NY has been going since tuesday 5th May. Overall, we have been hearing about really impressive genomics. On the poster session today, I came across an interesting new bias correction method for ChIP-seq data. Georg Stricker from Julien Gagneur lab in Gene Center of the University of Munich (LMU) presented a poster in BOG15.
The new method is based on Generalized Additive Models (GAMs).

Quoting the poster authors:

it smoothly models the coverage tracks as piecewise polynomials and is able to extract the signal component from the data.

Several thumbs up points for this framework:

  • Yields corrected smooth coverage tracks (after subtracting your input) with confidence intervals.
  • Can combine multiple proteins and replicates producing just one ‘aggregate’ track.
  • You can include additional covariates to regress out (e.g. GC content)
  • Coverage tracks are smooth, corrected for library size, so you can combine several tracks for better visualisation (such as conditions, IPs, etc)
  • GAM_ChIP

    Also really cool is that using smoothed coverage tracks allows for a simple way to find peaks based on the function and its derivatives.

    Worth watching out for this one when it is out!

    Advertisements

    Leave a Reply

    Fill in your details below or click an icon to log in:

    WordPress.com Logo

    You are commenting using your WordPress.com account. Log Out / Change )

    Twitter picture

    You are commenting using your Twitter account. Log Out / Change )

    Facebook photo

    You are commenting using your Facebook account. Log Out / Change )

    Google+ photo

    You are commenting using your Google+ account. Log Out / Change )

    Connecting to %s