Plot produce with sig. mutations at 0

[natir]

Hi,

We use SigProfiler in our cancer genome pipeline, on some sample decomposition verbose mode produce this output:

                                        ################ Sample 1 #################
############################# Initial Composition ####################################
   SBS1  SBS5  SBS96A
0   1.0   1.0   68.0
L2%:  0.055660557522576835
############################## Composition After Initial Remove ###############################
   SBS1  SBS5  SBS96C
0   0.0   0.0   70.0
L2%:  0.05906191827907873

Our sample match with SBS96C but figure produce contains graph for SBS1 and SBS5 this creates confusion. moreover, the number of sig. mutations signatures indicated is 0.

Code use to produce this figure:

    signatures = (
        output_dir
        + "SBS96/Suggested_Solution/De_Novo_Solution/De_Novo_Solution_Signatures_SBS96.txt"
    )
    activities = (
        output_dir
        + "SBS96/Suggested_Solution/De_Novo_Solution/De_Novo_Solution_Activities_SBS96.txt"
    )
    samples = output_dir + "SBS96/Samples.txt"
    output = output_dir + "Deconvolution_SB96_DeNovo"
    decomp.decompose(
        signatures, activities, samples, output, genome_build="GRCh38", verbose=True
    )

[mdbarnesUCSD]

Hi @natir, could you please provide your log file as well as the inputs used to generate the figures?

Thanks!

[natir]

Hi @mdbarnesUCSD,

We work with real human medical data so we can't provide VCF, but only mutational matrix, is this enough?

Thanks

[mdbarnesUCSD]

Yes, that would be great!

[mdbarnesUCSD]

Please reopen this issue if you are still encountering this problem.

[natir]

Hello,

Sorry to have taken so long to answer. We deal with a lot of samples, I don't have the data of the original problem anymore but I managed to keep the data of another similar case.

Which file do you exactly need ?

[natir]

I can't reopen this issue.

[mdbarnesUCSD]

Please provide the following files so that I can reproduce the plot above:

signatures = (output_dir+ "SBS96/Suggested_Solution/De_Novo_Solution/De_Novo_Solution_Signatures_SBS96.txt") activities = (output_dir+ "SBS96/Suggested_Solution/De_Novo_Solution/De_Novo_Solution_Activities_SBS96.txt" ) samples = output_dir + "SBS96/Samples.txt"

[natir]

You can find all of this her https://drop.infini.fr/r/xHsmyAxFLm#8sQTh/rXU10b9lgYyerBtK85gBYd1h5qmkbCNQUigHM=

Thanks for your help

[mdbarnesUCSD]

Hi @natir,

This is actually default behavior. The tool tries to introduce signatures 1 and 5 in all samples. If the signatures are not assigned then they appear with 0 mutations attributed. If you have any more questions please reach out.

[natir]

How we can change this behavior ?

[mdbarnesUCSD]

We will be looking in the future to make updates to the decomposition module. We will look into modifying this behavior when we do so.

[oliverartz]

I have a similar question: I understand that SBS1 and SBS5 might appear in the decomposition plots although we have 0 mutations corresponding to those signatures. In that case, I would expect the contribution to the final signature to be 0%. I do, however, sometimes observe, that SBS1 or SBS5 contribute 0 mutations but around 2-3% to the final signature. Is there an explanation for that? Where do the 2-3% come from, if we have no mutations that contribute?

Thanks for your help and the development! Great tool!