NA Values Appearing After Log Transformation of array Data

Question

[frankhailey645]

Is it reasonable to perform log2 transformation normalization on array data from GEO before conducting differential analysis with limma? After performing log2 transformation, many NA values appeared. How should I handle these NA values (should I remove them)?

Answer 1

[Ming Tommy Tang]

First, check the raw data if those are real NAs to begin with. If so, you can remove them. Also see https://www.bioconductor.org/packages/devel/workflows/vignettes/RNAseq123/inst/doc/limmaWorkflow.html for array data, it probably has some normalization tool for that microarray.

[frankhailey645]

Thank u for your reply. If there were no NA values in the original expression profile, but NA values appeared after log transformation, should an alternative normalization method be chosen in this case or remove NAs directively?

[Ming Tommy Tang]

unless there are negative values, (log2 will create NaN not NA). you need to understand why log transformation generates NAs.