How to set parameters to run SigProfilerClusters for INDEL? about sigprofilerclusters

[ebergstr]

Hi,

Can you provide additional information on your project (i.e. average number of indels per sample)? Can you also provide the list of parameters that you are using to run the simulator?

Best

[YingYa]

Hi,

Average number of indels per sample is about 36000 (from 50 to 345000).

Parameters for simulator:
sigSim.SigProfilerSimulator(project=project, project_path=project_path, genome=GRCh38, contexts=ID, exome=False, simulations=100,
updating=False, bed_file=None, overlap=False, gender=female, seqInfo=False, chrom_based=True, seed_file=None,
spacing=1, noisePoisson=False, noiseUniform=0, cushion=100, region=None, vcf=False, mask=None)

Best

[ebergstr]

Hi,

We have not benchmarked against samples with this high of indel TMB, so the simulator is likely struggling to complete many of the high TMB samples. Within PCAWG, we have on average ~1,400 indels per sample with maximum of ~137,000, which should run through without any problems. However, if you also have a large number of samples, I would recommend batching your project into multiple jobs. With projects containing more than 500 samples, we typically simulate across batches to manually speed up the process.

Best

[YingYa]

Hi,

Thanks for your suggestion.

You mean I could batching my project into multiple simulator jobs firstly, then run cluster separately or use the merged simulator results to run cluster?

Best

[ebergstr]

Correct, we typically:
1) Separate the samples into smaller project sizes.
2) Run simulator on each batch
3) Run clusters on each batch.

This should help speed up the process and/or locate problematic samples easier.

Best