Meta-Analysis based on the Control of FDR:
Combining yeast ChIP-chip data sets

- Saumyadipta Pyne, Steve Skiena and Bruce Futcher

- Appendices (preprint)

=================================================


Input


Lists: H, S, L (see references below)

Parameters:

Combined FDR level 0.06 (i.e. alpha)

Call = 1 (i.e. at least one list is required to reject the null hypothesis in order to qualify as significant)

________________________________________________________


Transcription Factors°

- the significant intergenic regions (using the above arguments) listed§

- sorted in the ascending order of the thresholded product of p-values 

- displayed as Excel spreadsheets (.xls format)

 

The fields -

A) Biding specific to the intergenic region upstream of this *gene*

B-D) The raw binding p-values due to lists H, S, and L respectively

E) The thresholded product (W_alpha) of p-values 

F) The lists that contribute to the above product

G) The Swi4 percentile rank due to I, if known 

H) The MBF percentile rank due to I, if known

 

° These 9 TFs are due to the data set in S


§ Based on subsequent finer calculation of the cutoff (largest p-value with
estimated q-value less than or equal to alpha'), the above updated lists
may differ by 0 to ±2 significant intergenic regions from earlier reports.

________________________________________________________


Statistics

- number of significant intergenic regions for Call = 3, 2 and 1 (and alpha = 0.06)

TFs All three lists At least 2 lists At least 1 list
Ace2 31 53 100
Fkh1 13 23 125
Fkh2 24 51 124
Mbp1 32 69 136
Mcm1 21 38 94
Ndd1 32 60 109
Swi4 65 106* 171
Swi5 35 78 162
Swi6 25 95 177

*67 out of 109 have entries in the list I with rank correlation | r | = 0.71 with Swi4 and 0.02 with MBF.  

Figure 2 (Swi4) in the paper

________________________________________________________

 

References : ChIP-chip experiments (TF local binding in S. cerevisiae) –

Abstract/Poster (RECOMB 2005)

Maintained by - spyne@cs.sunysb.edu