SUNY Albany RIP-Chip Analysis

CEL files for each sample were processed using the apt-probeset-summarize binary 
provided by Affymetrix. One qualitative (DABG) and two quantitative (RMA, PLIER) 
measures of expression were derived for each gene-level metaprobeset on the chips 
using the command line parameters given below: 

apt-probeset-summarize -a rma-bg,pm-gcbg,med-polish -a pm-gcbg,plier -a pm-only,dabg.neglog10=true  \
--precision 10 -p lib/HuEx-1_0-st-v2.r2.pgf -c lib/HuEx-1_0-st-v2.r2.clf -b lib/HuEx-1_0-st-v2.r2.antigenomic.bgp \
-m lib/HuEx-1_0-st-v2.r2.dt1.hg18.full.mps -o gene/ --cel-files cel_list

For a detailed explanation of the parameters please see Affymetrix documentation for 
apt-probeset-summarize:

http://www.affymetrix.com/support/developer/powertools/changelog/apt-probeset-summarize.html

An R session was used to import the expression values and experiment descriptions 
so that the following steps could be performed.

1. A conservative DABG p-value cutoff for presence was established at p < 1e-256 
(neglog10 value = 256). Visual inspection of DABG score distribution across all 
HuEx10ST chips bear this out as a natural bipartite division point for the data.

2. Probesets at or below the DABG p-val cutoff for all replicates were marked as 
present in the given condition & cell-line.

3. Lists of core probesets present in IP, in Total, and in Either are generated.
	   
4. Average log2 fold-changes for IP over total are taken for both the rma and 
plier expression scores.
	       
5. P-values for significant overexpression in IP replicates over total replicates 
are taken for rma and plier scores.
	          
6. The minimum of the rma & plier fold-changes is used as the overall fold-change.
		     
7. The maximum of the rma & plier p-values is used as the overall overexpression p-value.
		        
8. A default score of -10 is set and for all probesets present in IP, scores are 
replaced by -log10(max.p-value)*(min.log2.fold-change) [ranging from ~-.5 to ~15]
			   
9. Scores are scaled according to: x = 10*(x+10)+1 - now absent in IP probesets 
have a score of 1 and present in IP probesets range from ~100 to ~1000.
			     
10. A broadPeak bed file is created corresponding to the core gene-level probesets 
for each cell-line with the score field containing our scaled score, the signal 
field containing our minimum log2 fold-change, and the p-value field containing -log10 
of our maximum p-value. The q-value is left at the default of -1.