Question about Meso

[disulfidebond]

Hello, I'm trying to use SigMA to detect Signature 3 in mesothelioma samples. I followed the steps in the Wiki, but got somewhat confusing results. It reported no presence of signature 3 in all samples I provided to it, even in those our lab strongly suspect have HR deficiency.

When I tried to train SigMA using a subset of our WES data, after sending the message to the console "determining SBS scale for quick simulation", it consistently threw the error

Error in !m_ref$is_sig3 : invalid argument type

I did a bit of checking, and it looks like the matrix that is loaded in the get_inter_tt_suppl() function has all NA values for meso:

tmp <- system.file("ext data/matrices/matrices_96dim.rda", package="SigMA")
load(tmp)
matrices_96dim$matched_normal$tcga$meso$is_sig3
# output (truncated)
[1] NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA

When I ran the function in the debugger with a breakpoint at line 155 of the quick_simulation.R file, I got this output:

Browse[2]> sum(m_ref$is_sig3)
[1] 0
Browse[2]> sum(!m_ref$is_sig3)
Error in !m_ref$is_sig3 : invalid argument type

But when I loaded the example breast cancer file instead and ran quick_simulation(), I got this output in the debugger:

Browse[4]> sum(m_ref$is_sig3)
[1] 189
Browse[4]> sum(!m_ref$is_sig3)
[1] 482

Does this mean that the model for SigMA was trained with no Signature 3 for mesothelioma, and is this the correct approach to use?

Thanks!