I'm trying to identify polymorphisms in three samples. I'm usng Varscan and the default p-value threshold for calling variants is 0.99. If I want to get a false discovery call of 1/10.000 or better, should I set the p-value threshold as 0.00001? Deffault values seem too high... am I wrong?
Yes, the p-value can and probably should be set to a specific value based on your own threshold criteria. If you create a job that runs out of memory, you can make this less strict or consider moving to a local or cloud Galaxy with resources that you can control (assuming the job is not so large that it will not run anywhere - in Galaxy or not).
This is explained in the tool manual here: http://varscan.sourceforge.net/support-faq.html#output-p-value
For this tool and most, the underlying tool inside of Galaxy is wrapped and the usage within and outside of Galaxy is the same. Only where to actually set the inputs and other parameters differ. Meaning: options given on the tool form versus options given on the command-line are the same.
Hope this helps! Jen, Galaxy team