Hi,
I am new to proteomics and I see it is very popular to perform imputation (to replace missing data). There are several methods aiming to replace a missing value using the other replicates (by calculating the media, a probability...). I am quite reluctant to do that because I feel it is a bit like cheating although the publications seem to agree that it performs well.
What I want to do on the other hand is to account for missing values in all replicates in a particular condition. From my experiment, I know some of these proteins are actually absent. Do you know of any method to replace NA with something else ? I tried replacing all NAs with zero but I am afraid proteins just bellow the detection point in some conditions and just over in some other will be very significantly differentially expressed. I also tried modifiing by changing to a bit les than the minimum value of all the samples. It is a bit better but then I do the test on samples that have a variance of 0 and an other that has a high variance.
Any suggestions ? Experience ?
Thanks
