Dear colleagues, Thank you to everyone who sent information on the patch for the 64-bit version. It appears that my problem was much simpler: I used to run MB in a non-parallel version some years ago. Since I have moved to another institution a number of things have changed but I never read the new manual (ooops) or updated my input files accordingly. So here's the fix: If you use MB that is compiled as a parallel version it is set to run 2 runs by default. In this case, when specifying sumt do NOT include the .t in the file name as MB will add .run1.t / .run2.t and so on. When recalling the tree file MB cannot find the file if you already specified .t as the file ending. The sump command was ok (I could open the .p files in Tracer) but for continuity reasons I left it out now and is seems to be ok. Looks like I had a blond moment after all! Take care, Birgit On 21/5/09 15:16, "Birgit Meldal" wrote: Dear colleagues, I am having trouble getting MrBayes saving the consensus file from the sumt command. There is no indication that it shouldn't work automatically, however... I found a similar posting in the archives where it was pointed out that it was due to compiling it as a 64-bit binary. I don't compile my own program as it is running on the bioinfomatics server. I would be surprised no-one else has this problem as I know MrBayes is very well used on the server. It used to work at my former institution in an older version but not now. If anyone can enlighten me so I can save time but not doing it manually in PAUP that would be greatly appreciated! With kind regards, Birgit PS: Maybe I am just having a blond moment... Birgit Meldal, Ph.D. Division of Transfusion Medicine Department of Haematology University of Cambridge NHS Blood & Transplant, Cambridge Centre Long Road Cambridge CB2 0PT U.K. +44 1223 548049 bhmm2@cam.ac.uk