1. Separate the probes into "+" and "-" files

2. For each file, sort the probes based on genes (expression before median polish)

Gene probe +/- SuspCell Leaf Panicle Root

LOC_Os10g20970.1|12010.m05117|cDNA AGCAAAAATCATCAGCCTCTTCGTTACGATCGGTGA - 161.88889 212.22223 177.77777 178.0

LOC_Os10g20970.1|12010.m05117|cDNA TATCAAAAAATGACGGAGGGAAACACATCTCAAGTT - 409.66666 256.22223 270.66666 310.8889

..

3. Do median polish for each gene

a) Get the expression levels for one gene, with each line corresponding to one probe

e.g. try.data

b). Conduct median polish for that gene. Direct the output to a file

> a<-read.table("try.data",header=TRUE)

> sink(“output”)

>medpolish(a)

>sink()

e.g. output

Process output to get the expression pattern for the gene

Gene +/- Overall Nodule Root Stem Leaf Flower Seed

Name1 + 9.941 -0.0020 0.0080 -0.0525 0.0250 -0.0230 0.0075

c) Repeat a-b for every gene

Gene +/- Overall Nodule Root Stem Leaf Flower Seed

Name1 + 9.941 -0.0020 0.0080 -0.0525 0.0250 -0.0230 0.0075

Name2 + 10.36 0.16 0.55 0.87 -0.16 -0.41 -0.38

Name3 + 10.23 0.09 0.01 0.17 -0.01 -0.23 -0.19

…

d). Invoke R from command line to run a batch file (in window command console)

C:\Program Files\R\R-2.6.2\bin>R CMD BATCH mycmd

Where mycmd is a file containing:

a<-read.table("try.data",header=TRUE)

sink("output")

medpolish(a)

sink()

where try.data is file in C:\Program Files\R\R-2.6.2\bin. Output is the output.

4. Submit the files you get in step 2 (one for "+" and one for "-") and step 4c(one for "+" and one for "-").