Why only C > A,T,G and T > A,G,C mutations typically considered in 96 SBS context plots

Question

[BioStar22]

... even though all 12 possible mutations could theoretically be analyzed? Are purin base mutations redundant in a way to these pyrimidin base mutations or why are they not considered/ how are they count? I’m probably just missing something obvious here, but I’m a bit stuck and would really appreciate some help! Apologies if this is a very simple question — any clarification would be greatly appreciated. Thanks a lot in advance!

[GenoMax]

Looked up what SBS context plots mean. Some information here: https://cancer.sanger.ac.uk/signatures/sbs/

Answer 1

[jared.andrews07]

Because the signatures always have the mutated base represented by the pyrimidine base of the base pair. As far as I can tell, the original paper doesn't provide a clear rational for this, but I expect it's because it allows for one signature associated with a given mechanism or such rather than two.

[i.sudbery]

To clarify this, if its not obvious from jared.andrews07 answer - a C>A mutation is the same mutation as a G>T mutation, but on the opposite strand.

[BioStar22]

Thank you, both, that makes sense! It implies that mutations are always present on both strands = are fixated via repair/replication - is this correct?

[i.sudbery]

It does imply that, yes. It does not necessarily make it true though. Any mutation that has been replicated will be make a mutation present on both strands. Unless you are doing single cell sequencing, then the chance of detecting something that hasn't been repair/replicated is very low - each cell contains 4 DNA strands (two strand for each pair of chromosomes). If only one of those four strands is damaged, and this is only one cell in possible 10s of millions that contributed DNA to the sample, the chance of detecting is very low. Or, to put it another way, any mutation that is found in more than a single copy is almost certainly present on both strands.

However, DNA damage processes may leave mismatched, abradant or abasic sites where the damage is on only one strand, and these could be read as changes by a sequencing process. We would generally call these DNA damage, rather than mutations, but I don't really know how a sequencing machine would distinguish the two.

[BioStar22]

thank you