CNLOH when including variants with high non-genotyped cells

[EspressoKris]

Hi and thanks for the great tool,

I noted that when including somatic variants with high % of non-genotyped cells I get CNLOH due to the reduction in coverage that comes with it. Is there any parameter to take this into account without having to discard the variant?

Thanks

[e-sollier]

Hello, you can try using these command line arguments:

• --lohcost (default: 85.0). Setting a higher value (for example --lohcost 150) will increase the cost of LOH events, so you should see fewer of such events in the inferred tree.
• --dropoutrate (default: 0.05). Prior dropout rate. The dropout rate for each variant is inferred from the data, but if you use a higher value for the prior, then the inferred dropout rate should be higher. If the dropout rate is higher, then you don't need LOH events to explain many cells with read counts from only one allele.
• --dropoutrate_concentration (default: 100). Concentration parameter for the beta distribution of the dropout rate prior. Setting a higher value will force the inferred dropout rate to be closer to the prior mean (defined with the previous parameter), even if there is a lot of evidence (i.e. many cells) suggesting that the dropout rate should be different.