I am trying to understand the plots shown at COSMIC's website where they show mutational signatures taking into account whether mutations happen in the "transcribed" or "untranscribed" strand (see here for an example). As of right now, this makes no sense to me as to the best of my knowledge, from standard whole genome or exome sequencing it is impossible to determine in what strand the mutation happens. What you see is a mutation in both strands and you don't know which strand mutated first and which strand changed after to "stabilize" the first mutation.
When counting mutations without dealing with strand bias, I would simply get each mutation (always called in the reference strand) and see in what strand the gene that is hit is located to know what to count. For instance, say the mutation is C>A. If the gene is also in the reference strand I count +1 C>A or +1 G>T if the gene is in the anti-reference strand. However, I do not understand what would be different when accounting for transcription strand bias or how it is even possible to say for sure if something has occurred in the untranscribed strand.
