Empty optimalClusterSolution.txt

[d88020]

Hi.
I have a question about TitanCNA result.
I got results of 11 WGS data using snakemake.
But, in only 1 sample, optimalClusterSolution.txt was empty.
Only column names are there.
Also, I found out NaN values in tumor_sample_1_cluster1.params.txt file.
Could I get some advice on this?

[vleblanc]

I'm running into the same issue. At a ploidy of 2 or 4, every solution (1-5 clusters) has S_Dbw (LogRatio) values of NaN, although there are (AllelicRatio) values. At a ploidy of 3, the S_Dbw validity index (LogRatio) is quite high (~3) for most solutions, but for the 5-cluster solution it's 0.0 - I'm wondering if I should go ahead with this solution, or if this is indicative of larger issues with the sample itself. All of the solutions don't seem to indicate any CNVs (all segments are copy number 2 in the ploidy 2 solutions, and copy number 4 in the ploidy 3 and 4 solutions), so I'm wondering if there's a chance that the purity is 0 and that's why I'm having issues (I wouldn't expect that based on the sample, but I've also observed odd results in matched RNA-seq data).

[gavinha]

Hi @d88020 and @vleblanc ,

You're right that it is possible that TITAN is estimating zero tumor purity and that would potentially lead to NaN values. Hopefully, the optimal solution is the one that has nearly 0 purity. If not, then you would need to select this manually.

I had recently run into this issue myself where many of my samples had near 0 purity and I adjusted the solution selection to account for this. Sorry for the inconvenience if it is not working as expected for you.

Best,
Gavin

[vleblanc]

Hi @gavinha,

Thank you for the response! I do have another sample that I expected to have very low purity and it's being called as 0.3%, so in that case it's working as it should I think. It's also working well for the other 22 samples I have, so thank you for creating this helpful tool.

Regarding the problematic sample, I do have three other samples from the same tumour and the best solutions for these all have ploidy 3. All of the solutions for the problematic sample under ploidy 3 have non-NaN values, so I think one of these may be the best solution. However, is it normal to see a validity index of 0.0, or would you recommend I choose from the ones that have an index > 0? I know this is a lot of information so I'm happy to include figures/parameter tables if that helps.

Best,
Veronique

[serpei]

Hi @gavinha,
I stumbled upon the same issue. I'm using the nextflow pipeline and I got an error in the last step TitanCNA (lack of an input file), so it does not create optimalClusterSolution folder and crashes. Looking at files created in the previous step I found that optimalClusterSolution.txt was empty and I found out NaN values in tumor_sample_1_cluster1.params.txt. I would like to have summary plots and other results, so what do you suggest to do? Can I "create" manually an optimalClusterSolution.txt and relaunch the cextflow pipeline? And how can I select the optimal solution? Is there any manual explaining output files the pipeline produce?
Thank you,
Serena