Dear, 

I have fluidigm data which is qPCR values. the data is 27 x 96 table format. It has 3 time point(1, 10, 20), Each time point has 9 replicates. So 3x9=27 rows in the table.  96 genes are in columns in the table. 

I would like to get p-values with proper t.test. 

I am confused whether I have to use normal paired t.test or non-parametric wilcoxon sign rank test for this paired sample data. All data is from same sample with time intervals.

I have done shapiro test each column(96genes). 25 cases has bigger than 0.05 p-values in shapiro test. 71 cases have lower than 0.05 p-value. So They are mixture of normally-distributed and non-normally-distributed columns. I think Affymetrix deals with this issue with normalization with values and then t.test.  

The values in table are Et values(40 - Ct_value). 

Any advices would be welcome. 

I would appreciate it. 

Personally I use the limma package to analyse fluidigm data. It can directly analyze a matrix of Et values. You can do this directly using the limma package, or indirectly using the HTqPCR package.