Pindel: direct tumor/tumor comparison

Question

[volvox]

I've got a dataset with normal and two tumor samples (from the same tumor) for each patient. Let's call those n, t1, and t2. I want to compare insertions and deletions in t1 to those in t2. Is it advisable to run Pindel using directly t1 and t2, i.e., use t1 as normal sample? The alternative is to run n/t1, and n/t2 seperately; then compare two results. The latter would take twice as long. Also there is extra work required to compare the ouput files from n/t1 and n/t2.