Mpeak |
|
||
|
Mpeak method, which is a model-based method, is used to identify
peaks in ChIP-on-chip data. It
assumes that the sonication process approximately follows a Poisson Process.
Therefore, the log signal ratio should follow a triangular shape around a
true transcription factor binding site. Some typical examples are:
Triangles are fitted to every probe in the data by the
double-regression algorithm. For each probe, the best fitted triangle are
retained. By the assumption that in a small neighborhood of a true binding
site there should be no other binding sites, inhibition method is used to
identify the true peak. Here are some results:
(The first row is the original signals, the second row is the fitted
signals, and the third is the peaks identified by the program.) Multiple peak in one cluster can also be treated by Mpeak:
Download version 2.0 of the program (ReadMe file
included) What is new in the latest version? Need more information or free consulting? Please cite the following paper if you used Mpeak for the data analysis: Zheng, M., Barrera, L. O., B. Ren, and Wu, Y. N. (2007) ChIP-chip: data, model, and analysis. Biometrics, Vol 63, 787-796. Other Reference: Kim, T.H., Barrera, L.O., Zheng, M., Qu, C., Singer, M.A., Richmand, T.A.,Wu, Y., Green, R.D., and Ren, B. (2005) A high-resolution map of active promoters in the human genome. Nature, Vol 436, pp 876-880. |
