Yes, the exact same datasets work fine on another machine. We compiled 3.4.0 from source (our sys admin often prefers to do this when the path might be very different from what a software designer might expect). Ummm ... I didn't get that one. What path? The MIRA binary is (on Linux) completely static and totally self-contained. Yes, which is great. I think that perhaps our admin has gotten burned before with non-static binaries and prefers not to take that chance. I should mention that we have also had some difficulties in the past on that machine with binaries giving seg faults which we fixed by compiling from source. We are currently testing 3.9.5 and so far have not seen any segmentation faults (but the assemblies are not yet complete). 3.9.5 is also a completely different beast from 3.4.x ... but if it works OK for you, so much the better. Yes, given that we are assembling Ion Torrent data, we should probably be using 3.9.5 anyway. B.