Whilst Andreas Sottoriva and Trevor Graham's method for identifying neutrally evolving tumours, introduced in https://www.ncbi.nlm.nih.gov/pubmed/26780609, is still being debated, I wonder if there might be scope or interest in applying this ti the GLASS data for inclusion in the initial publication? As a preliminary inspection, I downloaded the mutation calls and, using the threshold suggested in the paper (0.12<=vaf<=0.24), applied the method to indicate which gliomas evolved neutrally in the lead up to, or following, treatment ( for each of the 201 pairs where data for the primary (TP) and first recurrence (R1) are available). This indicated that 41 primary tumours (~20%) and 41 recurrent tumours (~20%) met the requirement for neutrality (Rsquared>0.98) - and in 12 patients both their primary and recurrent tumour were labelled as neutrally evolving by this method. In the paper above it was shown that no GBMs met the requirement to indicate neutral evolution - but there were only 4 included. This would be a more thorough inspection and though would require further analysis: 1) Might be necessary to do the analyses in more depth (as per the original publication) to see, for example, how copy number affects the findings 2) The group have since shown that multi-region sampling is a better indicator of whether the tumour as a whole is evolving neutrally - are there sufficient multi-region samples in this dataset to inspect this? 3) Do neutrally evolving primaries do better/worse in terms of PFS or OS? etc etc But could potentially spark interesting discussion: 1) Neutral evolution indicates that the tumour cells have 'everything they need' to exhibit their phenotype - in these cases would it only be effective to therapeutically target clonal mutations because subclonal ones, even if usually though to be driver mutations, are not causing a selective advantage (i.e. this could be relevant for clinical trials for new targeted therapies: if the patient's tumour was shown to be neutrally evolving, would/should the presence of the mutation targeted by the trial drug be sufficient for that patent's inclusion IF the targetted mutation was found to be subclonal?) 2) If we have neutrally evolving primary tumours but non-neutral recurrences, are these the best pairs to inspect to identify candidate treatment-resistance conferring mutations (.e. those which dominate post-treatment because they were selected for during therapy specifically)? 3) Conversely, in cases where the primary is not evolving neutrally but the recurrence is, does that indicate that the clonal mutations in the recurrence are the best candidates for conferring resistance as a potential selective sweep caused by treatment has left the tumour with all it needs to continue; accumulating mutations linearly with no selective (dis)advantage? Just some food for thought...