Inconsistent result when validating

[lys1001s]

Hi, Thank you for making this tool!

I've tried validating msisensor2 tumor-only, WES mode. Original result obtained by msisensor paired mode is like below:

• Total number of sites: 200,043
• Number of somatic sites: 17
• % : 0.01

However, the result obtained by msisensor2 tumor-only mode with the same sample used for tumor sample above is like below:

• Total number of sites: 1,227
• Number of somatic sites: 29
• % : 2.36

Is it because msisensor2 was trained with ~3000 site models? Then what do I need to adjust so that I can have similar % msi with the first result?

Thank you