PyClone 0.13.0 ZeroDivisionError: division by zero

[David Zhang]

Hi Andrew,

I was running PyClone 0.13.0 for mutations from four samples (mice) by which about 6000 mutations are common in a total of 10,000 mutations. Then I got the below error after 10000 iterations:

 

Iteration: 9900
Number of clusters: 13
DP concentration: 1.54672472032
Beta-Binomial precision: 437.505977442


Traceback (most recent call last):
  File "/opt/anaconda3/envs/pyclone/bin/PyClone", line 11, in <module>
    load_entry_point('PyClone==0.13.1', 'console_scripts', 'PyClone')()
  File "/opt/anaconda3/envs/pyclone/lib/python2.7/site-packages/pyclone/cli.py", line 78, in main
    args.func(args)
  File "/opt/anaconda3/envs/pyclone/lib/python2.7/site-packages/pyclone/run.py", line 63, in run_analysis_pipeline
    thin=args.thin
  File "/opt/anaconda3/envs/pyclone/lib/python2.7/site-packages/pyclone/run.py", line 359, in _build_table
    thin=thin,
  File "/opt/anaconda3/envs/pyclone/lib/python2.7/site-packages/pyclone/post_process/clusters.py", line 48, in load_summary_table
    thin=thin,
  File "/opt/anaconda3/envs/pyclone/lib/python2.7/site-packages/pyclone/post_process/clusters.py", line 112, in load_table
    cluster_sample_posterior = _compute_posterior(cluster_sample_data, density, mesh_size)
  File "/opt/anaconda3/envs/pyclone/lib/python2.7/site-packages/pyclone/post_process/clusters.py", line 140, in _compute_posterior
    posterior[cellular_prevalence] += density.log_p(data_point, BetaData(cellular_prevalence))
  File "/opt/anaconda3/envs/pyclone/lib/python2.7/site-packages/pyclone/pyclone_beta_binomial.py", line 113, in log_p
    return self._log_p(data, params)
  File "/opt/anaconda3/envs/pyclone/lib/python2.7/site-packages/pyclone/pyclone_beta_binomial.py", line 121, in _log_p
    params.x, data.tumour_content, self.params.x
ZeroDivisionError: division by zero

I installed the PyClone 0.13.x using "Sandboxing python" and was able to run the example data set successfully.

Do you have some suggestions to fix it?

Thanks,

David

[Andrew]

Hi David,

So it looks like the clustering part went well, but the post-processing is failing. I assume you are running with the pipeline command. Could you try running the `build_table`, `plot_clusters` and `plot_loci` command separately. You should not need to rerun `run_analysis` since the output should still be around. If you need help with the commands run `PyClone build_table -h`, for example.

If you could send the trace file over, I can also take a look.

Best wishes,

Andy