Closed AngelGarciaDeLaTorreGarcia closed 10 months ago
Hi @AngelGarciaDeLaTorreGarcia,
Can you check the ASPCF plot to make sure that the segmentation went fine? The logR track looks pretty spiky to me so it may have been over-segmented. Was it corrected for GC% and RT? If you used the default penalty (70), you may want to try a higher value such as 100, 140 or 200. Also, there is a blob in the BAF track for chr2, I'm not sure ASCAT identified an allelic imbalance there since the signal isn't very clear.
Checking all steps, plots and QC will reduce the risk of wrong calls but it's all about predictions unless you have experimental ploidy validation.
Cheers,
Tom.
Hello,
Yes, it was corrected for GC% and RT and this is my ASPCF plot (I don't see remarkable differences with the rest of the samples processed).
I may try to change the penalty.
Thank you very much,
Ángel
EDITED: My apologies; I have attached one of the documents with which I was making comparisons.
Hi @AngelGarciaDeLaTorreGarcia,
The segmentation isn't too bad actually, it isn't as spiky as I would have thought. ASCAT misses the BAF segments on chr2 but the rest looks fine. Maybe you can try updating ASCAT: we made some changes to the BAF segmentation earlier this year (v3.1.2), this might help.
Ultimately, I can think of two tweaks:
ascat.bc$Tumor_BAF_segmented[[1]]
object to manually set the segmentation to the right baseline (after ascat.aspcf
and before ascat.runASCAT
).ascat.output$distance_matrix[[1]]
object to find the best purity/ploidy fit (lowest value in the matrix) and force ASCAT with the rho_manual
(=purity) and psi_manual
(=ploidy) parameters from the ascat.runASCAT
function. This is the only way to get ASCAT profiles with goodness of fit <80% (otherwise ASCAT says it cannot find a fit), then compare the ASPCF plot and the CNA profile to assess whether the profile makes sense.I would do tweak 1 (the easiest) since this does not force any fit. If you're interested in tweak 2, I recommend you also do tweak 1 (hack ascat.bc$Tumor_BAF_segmented[[1]]
and explore ascat.output$distance_matrix[[1]]
).
Cheers,
Tom.
Hi!
I encountered a similar issue as previously reported. During the initial attempt, ASCAT failed to determine an optimal ploidy. However, on a subsequent trial with extensive min_ploidy and max_ploidy ranges, ASCAT computed a ploidy of 25 for my sample, which is impossible.
I have checked the plots generated and the mean average of coverage per chromosome and they seem fine.
The first image is LogR and the second is BAF.
How do I make sure the sample is okay for calculating ploidy? Are there other things I should check to figure out what's going on?
Thank you in advance,
Ángel