Table I. The groups participating in CAFASP-1

PROGRAMS	url	REF	Comments
3D-PSSM and (1D+3D)-PSSM	http://bonsai.lif.icnet.uk/ foldfitnew/index.html	[19]	Two methods, both accessible from the same url.
BASIC	http://cape6.scripps.edu/ leszek/genome	n.a.	The server is being moved to the following address: bioinformatics. burnham-inst.org
frsvr_SDP, frsvr_SDPM, frsvr_SDPMA2	http://www.doe-mbi.ucla.edu/ people/frsvr/submit.html and http://www.cs.bgu.ac.il/~bioinbgu	[8]	Three methods run for each submission
GenTHREADER	http://globin.bio.warwick.ac.uk/ psipred	[4]
Karplus1, Karplus2, Karplus3	http://www.cse.ucsc.edu/ research/compbio/HMM-apps /model-library-search.html or /T98-query.html or /model-library-search.html	[11] [20]	Three variants of the SAM-T98 method
pscan	http://www.biokemi.su.se/ ~server/pscan/profscan2_ga11.html	[21]
PSI-BLAST; BORK , NCBI	http://dove.embl-heidelberg.de/3D http://www.ncbi.nlm.nih.gov/ cgi-bin/BLAST/nph-psi_blast	[17]* [6]	Two servers running supposedly the same program
topits	http://www.embl-heidelberg.de/ predictprotein	[22]

* NOTE: The 3D server of the Bork group uses PSI-blast with an E-value threshold E=0.001. It is thus reliable, but also rather conservative in its predictions; it does not report homologies below the threshold. Thus its results are not comparable to those of the other servers that always report the top ranking hits, regardless of their scores. PSI-BLAST at NCBI does report below threshold hits, and like Bork's server, the only correct hits found were the 5 family-level targets plus target T0074.