Training a new model on panel data beyond Sig3

Hi!

We are interested in training a new model for a different gene panel; the guide helped a lot.

We want to detect other signatures beyond Sig3. I have read that we can use colname_truth_tag parameter in tune_new_gbm(). Default value is is_sig3, Which are the alternative values? is_sig4, is_sig5, etc?

Also, following the test_tune_example.R I observe that calling tune_new_gbm() is just part of the process. Before it, quick_simulation() is called. However, the comments say To tune a new model that fits the SNV count in our dataset we first simulate a new dataset from WGS data for which the sig3 is known from WGS analysis and is more reliable. If quick simulations are based on Sig3 data, Is this step limiting our ability to detect other signatures beyond Sig3?

Thanks a lot in advance

Hi @jpuntomarcos Are you eventaublly able to run this? I am running into the same issues. Thank you!