I am trying to recalculate the results of one series (GEO) but having trouble getting the same results , these are my steps :
- got raw data (.cel files)
- Normalized them using the method (Mas5)
- Log transformation
- Mean value of each group of samples:(f.e: gr1.mean,gr2.mean,..,control.mean)
- Calculating the ratio(according to paper): gr1cratio= gr1.mean-control.mean -gr2cratio =gr2.mean-control.mean
- 2^(ratio values)
Then I get (for example ) such a table :
gr1ratio gr2.ratio Gene1 2 14 Gene2 5 6 Gene3 2 2 Gene4 5 8
So they say in the paper they selected only the transcripts that are twofold up or down compared to control. Does it mean that when I have a 2 in the table that the fold change is 2 and therefor it twofold up? This would not match with the result they are presenting in their paper , and based on this there would not be values that are twofold down anyway
I tried to use their normalized data , they are providing for each sample separately on the GEO , but I again could not reach their results .
So what I am missing .. Do I have to compare the values of the table with log (mean value of the control group ) or maybe with the normalized(mean value of the control group) ?
Thank you very much !