Can I search a very small (1 or 2 protein) database?
No, you need to search the same FASTA database used in the previous analysis. The Scaffold database should consist of at least 100 proteins. X! Tandem depends on a large database to generate reliable results. So if you are planning on using X! Tandem bundled into Scaffold, you should use a large database. Mascot and SEQUEST also rely on a large database to generate reliable results. If you use a small database in a SEQUEST analysis the Xcorr values will become ridiculously low.