There are sites that attempt to do prediction even for new folds. The models are not good, though, unless there is a homolog in PDB. There were 98 servers registered for CASP7, 56 of which were producing 3D models, 7 of which were producing alignments. The other 35 were either entered in subcategories (doman prediction, contact prediction, ...) or were non-functional. I don't know how many of the 56 servers are actually free to the public with reasonable wait times. (One of the best at casp6 was Robetta, and there is a free Robetta server, but the queue is months long.) The one that currently looks best in CASP7 is the Zhang-Server, but I think that requires even more processors than Robetta and is not publically available (I may be wrong on that, as the Zhang-Server is a new entry to CASP.) I have a new server that is decent (better than my old servers and easier to use), but not top of the line any more. It does about as well as human-assisted prediction 2-4 years ago, but there are meta-servers doing somewhat better. Our server only uses one (rather ancient) CPU per job, so responses take 10-24 hours. This means that we can offer the service for free: http://www.soe.ucsc.edu/research/compbio/SAM_T06/T06-query.html In addition to the model, there are lot of other useful files provided: sequence logos, predictions of local structure, ... . ------------------------------------------------------------ Kevin Karplus karplus at soe.ucsc.edu http://www.soe.ucsc.edu/~karplus Professor of Biomolecular Engineering, University of California, Santa Cruz Undergraduate and Graduate Director, Bioinformatics (Senior member, IEEE) (Board of Directors & Chair of Education Committee, ISCB) life member (LAB, Adventure Cycling, American Youth Hostels) Effective Cycling Instructor #218-ck (lapsed) Affiliations for identification only.