On Sat, May 7, 2011 at 12:01 AM, Douglas Rhoads drhoads@uark.edu wrote:
We have a local install of Galaxy and are using it for training grad and undergrad students (using the Windshield Splatter data). We have a relatively new install and the Megablast seems to be doing something funny with the output in that column 2 which is supposed to be the GI of the database hit is now two numbers joined with an underline (see below). I need to tell my IT people what is wrong so they can correct the problem. BL1-1-3 157649053_111750 91.30 46 4 0 10 55 58541 58496 2e- 06 60.0 BL1-1-5 161788633_175843 100.00 26 0 0 15 40 48812 48787 4e- 04 52.0
Which version of BLAST do you have?
If you're using the "megablast" tool in Galaxy, then it will be NCBI "legacy" BLAST using the blastall binary.
If you're using the (commented out by default) BLAST+ wrappers, then it will be using the blastn binary. There were changes to the ID handling in the tabular output in BLAST 2.2.25+ as I recall.
Another important factor is how the BLAST database was made (with or without the -parse_seqids flag).
Peter
P.S. Questions like this about local installs might be better on the galaxy-dev list, CC'd.