search

VanLoo-lab / ascat

ASCAT R package
https://www.mdanderson.org/research/departments-labs-institutes/labs/van-loo-laboratory/resources.html#ASCAT
162 stars 85 forks source link

ASCAT could not find an optimal ploidy and purity value for sample #162

Closed mariaalexandrastanciu closed 10 months ago

mariaalexandrastanciu commented 10 months ago

Hi,

It seems that I ran into this problem for most of my WES samples(~80). I have the normal WES as well and the coverage is around 250 for tumor samples and 100 for normal samples.

I believe that the BAF is looking strange, but I don't know what can be the issue.

Here is an example after correction: Tumor LogR/BAF:

After_correction_NRH15001 tumour

and

Normal LogR/BAF: After_correction_NRH15001 germline

Can you give any suggestion on what is happening? I do have some samples for which I have purity/ploidy, but they as well look weird:

Tumor after correction: After_correction_NRH30008 tumour

Normal after correction: After_correction_NRH30008 germline

Thank you, Alexandra

tlesluyes commented 10 months ago

Hi @mariaalexandrastanciu,

For NRH15001, it seems like your BAF track corresponds to the logR track, can you double-check the parameter you use when running the ascat.loadData function?

The logR for normal will always be a flat line at 0, that's expected. I'm not sure what happens for your sample NRH30008 but the logR is so noisy that it should be discarded. Does it have a lower coverage (or its matched normal) or anything particular? Is the reference genome (hg19/38) the same for your entire batch?

Cheers,

Tom.

mariaalexandrastanciu commented 10 months ago

Hi Tom,

Thank you very much for your fast reply. I was indeed messing with loadData parameters.

For the NRH30008 sample, I do have lower coverage, 35x, compared with most of my samples. Do you have any suggestion on running ASCAT on this sample, or you would advise to discard it ?

Bests, Alexandra

tlesluyes commented 10 months ago

Hi @mariaalexandrastanciu,

Could you please share the ASPCF plot for this sample? It could be the segmentation went fine (even though the raw data is extremely noisy). If it's also very noisy, I recommend you try our ASCAT.sc tool (https://github.com/VanLoo-lab/ASCAT.sc) which leverages off-target reads to derive copy-numbers. This may help in getting a clean CNA profile.

Cheers,

Tom.

mariaalexandrastanciu commented 10 months ago

Hi,

Here it is:

NRH30008 ASPCF

Bests, Alexandra

tlesluyes commented 10 months ago

Hi @mariaalexandrastanciu,

Yep, it's very noisy and there aren't many options I'm afraid. ASCAT.sc may be able to help here, but you should not consider the output from ASCAT.

Cheers,

Tom.

mariaalexandrastanciu commented 10 months ago

Ok, I see.

Thanks a lot for your input!

Bests, Alexandra