I've checked in the changes. Looking back at my notes I can see why I left out MSF and Clustal from the auto-guessing method. By adding them we end up treating some alignment formats differently to others i.e. MSF/Clustal vs. fasta/raw. As fasta and raw can be either alignments or single sequences, depending on the intention of the user I thought is better never to allow guessing and always make the application programmer specify explicitly themselves (using the method I mentioned in my other post).
However, I think I was wrong as it's obviously given a nasty surprise. So I've applied the fix and added the relevant tests. Keith -- - Keith James <[EMAIL PROTECTED]> Microarray Facility, Team 65 - - The Wellcome Trust Sanger Institute, Hinxton, Cambridge, UK - _______________________________________________ Biojava-l mailing list - [EMAIL PROTECTED] http://biojava.org/mailman/listinfo/biojava-l
