I plan to do a differential gene expression using RT-qPCR of 10 genes on 3 strains cultured in different enviormental conditions of 3 factors (2 levels each one). In order to do so, I want to follow a DOE metodology. Since I think strain type can be considered as a factor itself, I end up with 4 factors, 1 at 3 levels (D) and 3 and 2 levels (A,B,C).
For this reason, I would like to use a Taguchi L16(2^15) design. To my knowledge, taguchi arrays can be further modified using different techniques to readjust factor levels. In my case, I introduce D factor in columns 8 and 7 leaving free the colum where the interaction between 7 and 8 is allocated. conbination of 2 + 1 is now 3rd level. Then, other factors and the interactions are assigned accordingly:
Therefore, the original L16(2^15) matrix includes the 3-level factor (D) and 14 columns
However, the resulting matrix is not ortogonal. I calculated the loss of ortogonality< (NOA, Nearly Ortogonal Array), which results in a 1,60% loss. I rearranged some levels in D column to make the design more balanced resulting in a loss of ortogonality of 0.5%
Is it possible and conceptually fine assesing the effect of each factor and interactions between them on each gene normalized expression using this DOE approach? I would really appreciate a second opinion.