What are the stringent outputs and why do I have different calling results?

[ginnyintifa]

I performed lofreq somatic to identify somatic variants in a tumor sample T against a normal sample N1, expectedly I get a list of somatic variants. Additionally I get some other files, relaxed, and stringent etc. And these files are for both the normal sample and the tumor samples.

Then I performed the somatic calling again using the same tumor sample T against a different normal sample N2, I fount that the stringent file for the same tumor sample contains slightly different calling results (I understand and expect the somatic variants are different between the calls. but why is the stringent files --- presumably without the somatic calculation are different for the same tumor sample? ). I wonder why is that? Are there any randomness in the calling process?