On Tue, 24 Aug 2004, Goel, Manisha wrote: > > >> -----Original Message----- >> From: Goel, Manisha >> Sent: Tuesday, August 24, 2004 2:15 PM >> To: 'pdb-l at sdsc.edu' >> Subject: Consesus of Multiple sequence alignment >> >> Hi all, >> >> I am new to seq alignments and I am at present using a couple of >> programs to align my sequences (ClustalX, MUSCLE, T_Cofee) >> The alignment agrees in most parts but still differs substantially, Is >> there any automated method of combining the Multiple seq alignment >> from various sources to get a consensus alignment. Otherwise, I guess, >> I have to compare all of the outputs to generate one. I have heard about methods to build 'profiles' from each multiple alignment, and then compare the profiles, but this probably isn't what you need. Each different alignment program has its own strengths and weeknesses (not always in equal measure). Your best bet is trying to work out which program is giving you the most sensible (most stable) alignment. Try showing your alignments to an expert, or compare them to structural data, or try to predict secondary structure to validate the 'inserts' in your alignment. Perhaps you can get better results by combining all your data into one consensus consensus, but I guess you need to properly understand the components first? Best of luck! Dan. >> >> Thanks, >> - Manisha >