Tweet
A few months ago I have asked here whether FPKMs in general can be used like microarray expression levels. While Dr Anders has kindly gave his opinion (that some transformation (such as log) might be needed), I wonder if this holds true for any other kind of analyses that takes microarray expression as input?
Specifically, I am planning on performing Quantitative Trait Transcript (QTT) and/or WCGNA on these data, but I didn't see any discussion on their applicability anywhere.
Unfortunately, the papers involved in these methods seemed to be confined to microarray expressions.
On close inspection, I have noticed QTT basically relies on standard Pearson correlation between two linear regression models. I haven't completely reviewed about WCGNA's algorithm, but the adjacency is also based on Pearson correlation.
Of course Pearson's rho test is based on linearity. But does that make the usage of FPKMs in these methods inappropriate?
Thanks for the answer in advance!
Specifically, I am planning on performing Quantitative Trait Transcript (QTT) and/or WCGNA on these data, but I didn't see any discussion on their applicability anywhere.
Unfortunately, the papers involved in these methods seemed to be confined to microarray expressions.
On close inspection, I have noticed QTT basically relies on standard Pearson correlation between two linear regression models. I haven't completely reviewed about WCGNA's algorithm, but the adjacency is also based on Pearson correlation.
Of course Pearson's rho test is based on linearity. But does that make the usage of FPKMs in these methods inappropriate?
Thanks for the answer in advance!