Salmon - is there any way to induce output to enable a pileup for read depth across transcripts

[Matthew Hockin]

I suspect this is essentially impossible given the kmer mapping strategy used, but can you induce Salmon to generate a file that could be used to generate (or itself contains) a format like a pile up.  

I am aware of alternative strategies that will generate this type of data, in fact using them lead me to ask this question.  I have a data set that apparently has some 3' read bias and I can demonstrate this using other mapping tools and thus this is almost certainly also true in our Salmon output but its not something that I think I can "know" using Salmon.  Is this correct?  

Thanks-