Low purity but high VAF

[yavorska]

Hi there,

We were hoping to use CNAqc to detect errors in some of our purity estimates (particularly for low TMB cases). We've found a few samples where analyze_peaks() passes with a low estimated purity (~0.2).

However, based on the VAF plot this seems unlikely (there's a clear peak close to 0.5 in diploid regions suggesting a much higher purity).

Are there any settings that could be modified to change this behaviour? I've tried using the "rightmost" method but that seems to cause pretty much every estimate to fail.

Thank you!

[caravagn]

Hi, I see your panel D. I agree that it should be possible to select the 50% peak. Likely it's matter of tuning some peak-detection parameters, the defaults are set to work well with higher TMB and maybe this is poor for your specific case.

We can try to help you find better params, but it would help us to work on your VAFs. Is there any way you can share a synthetic copy of data from panel D? For instance, you can make up nucleotide positions, ref/alt etc -- we just need the same read count values.

[yavorska]

Hi @caravagn,
That would be great! At least one such sample is available in the research environment (https://www.genomicsengland.co.uk/research/research-environment). If you raise a GEL service desk ticket, I can point you to it, feel free to tag me directly, @Olena Yavorska.

[caravagn]

@AliceAntonello can you do open the ticket following @yavorska?

[AliceAntonello]

Hi @yavorska, I found the sample you referred to in the research environment, could you share with me the code you used to produce the CNAqc object, so that I follow the same steps as you?

[yavorska]

Hi @AliceAntonello, I emailed the details a while back to the account that was used to generate the SD ticket. Can you confirm if you got them?

[AliceAntonello]

Hello, Yes, I confirm I received it. I apologise for the delay in responding to your request, I’ll get back to it as soon as possible. Sorry again.

…

Il giorno 19 set 2023, alle ore 18:04, yavorska ***@***.***> ha scritto: Hi @AliceAntonello <https://github.com/AliceAntonello>, I emailed the details a while back to the account that was used to generate the SD ticket. Can you confirm if you got them? — Reply to this email directly, view it on GitHub <#26 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AVK5BSYXKCA7PPXF7FBMK3TX3G67ZANCNFSM6AAAAAA2RE6V5E>. You are receiving this because you were mentioned.

[caravagn]

Has this been resolved? If yes, @AliceAntonello you should close the issue.

[AliceAntonello]

Hi! I apologise but I was not able to solve the issue. I tried to tweak some parameters to see if I could change the result, but in this particular case the purity estimate is too off and the input purity too low. I think we would need to try another strategy to address this issue.

…

Il giorno 14 ott 2023, alle ore 09:40, Giulio Caravagna ***@***.***> ha scritto: Has this been resolved? If yes, @AliceAntonello <https://github.com/AliceAntonello> you should close the issue. — Reply to this email directly, view it on GitHub <#26 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AVK5BS45OEKPGXSE6AV6HGTX7I6X7ANCNFSM6AAAAAA2RE6V5E>. You are receiving this because you were mentioned.

[caravagn]

I think in general it is hard to believe that the peak close to 0.5 is actually a clonal peak. If these are WGS samples it seems plausible to find hundreds of clonal mutations, and not 3.

What tumours were this @yavorska ?

[yavorska]

Hi @caravagn, it's an ALL.