SigProfilerAssignment error / specifying DBS_activities error

[katieskinner98]

Hi here!

I am trying to run an analysis on a set of samples, which roughly half have a low DBS count (<20). All of these samples fail at the SigProfilerAssignment step. The code I am specifically using is:

if __name__ == "__main__":
    topography.runAnalyses(genome = genome, 
                           inputDir = input_dir,
                           outputDir = output_dir,
                           jobname = "sample",
                           numofSimulations = 5,
                           transcription_strand_bias = True)

The error I get is:

--- SigProfilerAssignment for DINUCs using cosmic fit
Removing samples with zero TMB ...... 
Assigning COSMIC sigs or Signature Database ...... 

ValueError: cannot convert float NaN to integer

As mentioned, these samples do have DBSs, and I have successfully run SigProfilerAssignment on them before. So, as a way to get around this issue, I specified the activities for these samples:

if __name__ == "__main__":
    topography.runAnalyses(genome = genome, 
                           inputDir = input_dir,
                           outputDir = output_dir,
                           jobname = ""sample,
                           numofSimulations = 5,
                           transcription_strand_bias = True,
                           dbs_activities = /path/to/Assignment_Solution_Activities.txt)

This fixes the above error, but introduces a new error:

--- Merge real mutations with mutation probabilities

--- Merge simulated mutations with mutation probabilities

--- Fill tables/dictionaries using real mutations
object address  : 0x17e866200
object refcount : 3
object type     : 0x10528bf98
object type name: TypeError
object repr     : TypeError("'NoneType' object is not subscriptable")
lost sys.stderr

I am not sure of the origin of this error or why it may be happening. If I run the second command, then re-run the first one (i.e., no longer specifying the "dbs_activities" option), the analysis runs fine as, I assume, the correct files have been generated for the SigProfilerAssignment to not error out, and the dbs_activities flag is not triggering the second error.

Please let me know if any additional information is needed. Thank you so much for your help!

Katie

[burcak-otlu]

Dear @katieskinner98,

Thank you very much for using SPT.
Could you please share your data and Python file with me so that I can replicate the issue on my side?
You can reach me at burcako[at]metu.edu.tr

Best wishes,
Burcak

[katieskinner98]

Hi Burcak,

Thanks for your response! I have emailed the data and python scripts to you just now. Please let me know if you did not receive it or if there is any other information you need from me!

Best wishes,

Katie

[burcak-otlu]

Hi Katie,

Please update SPT with version 1.0.108.

1. This updated version fills all matrices before the SPA call. Since SPT calls SPMG chromosome based there were, not all matrices were filled, but the chromosome-based files. SPT was providing the path to the chr1 matrix for the SPA call. For your case, for the provided sample vcf, there were no mutations on the chr1. And this was creating a problem for joblib during the SPA call. Now, SPT fills all matrices before the SPA call, and provides all matrices for the SPA call and the joblib problem is resolved.

2. SPT requires at least 2000 subs, 1000 indels, and 200 dinucs to run the analysis. We require as much input as possible to observe a topographic pattern. Since you have a low number of mutations, please try your data as follows:

load libraries

from SigProfilerTopography import Topography as topography

strand bias

if name == "main":
topography.runAnalyses(
# set arguments accordingly
transcription_strand_bias=True,
num_of_sbs_required=1,
num_of_dbs_required=1,
num_of_id_required=1
)

Please let me know if your problem is resolved.
Thank you for your patience.

Best wishes,
Burcak Otlu