Plot produce with sig. mutations at 0

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]

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!