If (heaven forbid) your O database is corrupted such that you cannot read it in at the start of a new session, don't give up hope! Start O from scratch and then do the following
O > sam_atom_in old_db.o
This will read in your database as far as it can. If the main point is to save a particular molecule or skeleton, then use Sam_at_out or Write_formatted. Then start the program again from scratch and carry on working.