Prevalence << 1 for clonal mutations in diploid regions...

[Davide Cittaro]

Hi all, 

First of all, I'm sorry if this question has been somehow already posted. However, we have FACS sorted samples at diagnosis and relapse, we ran sequenza to get CNA and purity estimation. We indeed confirm that purity is very high (in the 0.85-0.95 range) and it is set in the yaml config file. We are using parental_copy_number prior (we also tested major_copy_number) with beta_binomial model, 100k iterations. We observe a large fraction of supposedly clonal mutations during the disease progression, in particular variants in diploid regions which have similar VAF (close to 0.5) in all time points. We expected the prevalence to be close to 1 (due to normal CN, 0.5 VAF and high purity) but we observe a slighly lower one (0.75). How could this be? 

Thanks, 

d

[Andrew]

Hi David,

It is hard to know for certain. Try taking the lowest estimated cellularity values. If the estimated tumour content is to high then you do end up with clonal SNVs having lower prevalence. Also note, that there is typically some disagreement between genomic estimates, pathologists and FACs for cellularity.

Also if the coverage is on the order of 50x-500x there may be a lot of uncertainty. What is the std deviation of the estimate prevalences? Does it overlap with 1.0?

Cheers,

Andy