Is it possible to infer pathway activity from RPPA (reverse-phase protein array) data from for instance TCGA? I've read some papers where they say it is possible, but unfortunately they don't explain how.
Sure, just use any standard pathway analysis program. However, performing pathway analysis in this way is immediately biased by the fact that only certain genes were profiled via RPPA.
Check the input distribution of the RPPA data. While some pathway analysis tools may only take a list of genes / proteins, others may want, in addition, the expression data, which one would usually have following a normal distribution.