PFRMAT TS 
TARGET T0054 
AUTHOR 9070-5088-8627 
REMARK  
REMARK Prediction date: Tuesday June 23, 1998 
REMARK Group name: UCSC-compbio 
REMARK Authors: Christian Barrett, Melissa Cline, Mark Diekens, Kevin Karplus, 
REMARK 	 David Haussler and Richard Hughey 
REMARK University of California, Santa Cruz 
REMARK  
METHOD Overview 
METHOD  
METHOD Fold recognition was performed using the Target98 (SAM-T98) method 
METHOD [3] using SAM version 2.1.1 [1], a refinement of the methods developed 
METHOD by this group for CASP2 [2].  This method attempts to find and multiply  
METHOD align a set of homologs to a given sequence, then create an HMM from that  
METHOD multiple alignment. 
METHOD  
METHOD First, a set of sequence weights is determined from the alignment.  Next,  
METHOD Modelfromalign is used to build the model from the alignment and the  
METHOD sequence weights.  Finally, hmmscore performs a local, all-paths scoring  
METHOD of the sequences, using a reversed-sequence normalization feature. 
METHOD  
METHOD The weighting method, detailed in upcoming publications [3,4], 
METHOD combines the Henikoffs' scheme [5], Dirichlet mixtures [6], and an 
METHOD entropy method to set the final weights. 
METHOD  
METHOD Alignment generation 
METHOD  
METHOD The initial step uses BLASTP to search NRP twice: once to produce a set 
METHOD of very close homologs, and once to produce a set of possible homologs. 
METHOD  
METHOD The method then uses multiple iterations of a selection, training, and  
METHOD alignment procedure.  Each iteration involves an initial alignment, a set  
METHOD of search sequences, a threshold value, and a transition regularizer.  
METHOD  
METHOD The first iteration uses a single sequence (or seed alignment) as the  
METHOD initial alignment and the close homologs found by BLASTP are used as the  
METHOD search set.  The threshold is set very strictly, so that only good matches  
METHOD to the sequence are considered.  This iteration uses a transition regularizer  
METHOD that was designed to match the gap costs used by BLASTP. 
METHOD  
METHOD On subsequent iterations the input alignment is the output from the 
METHOD previous iteration, the search set is the larger set of possible 
METHOD homologs found by BLASTP, and the thresholds are gradually loosened. 
METHOD The second through second-from-last iteration use a ``long-match'' 
METHOD transition regularizer, and the final iteration uses a transition regularizer  
METHOD trained on FSSP alignments. 
METHOD  
METHOD References 
METHOD [1] R. Hughey and A. Krogh, CABIOS 12(2): 95-107, 1996. 
METHOD     http://www.cse.ucsc.edu/research/compbio/sam.html.   
METHOD [2] K. Karplus, K. Sjolander, C. Barrett, M. Cline, D. Haussler, R. 
METHOD     Hughey, L. Holm, and C. Sander, Proteins: Structure, Function, and  
METHOD     Genetics, Suppl. 1, 134-9, 1997. 
METHOD [3] K. Karplus, C. Barrett, and R. Hughey, Technical Report UCSC-CRL-98-06, 
METHOD     Department of Computer Engineering, Univ. of California, Santa Cruz, 1998. 
METHOD [4] J. Park, K. Karplus, C. Barrett, R. Hughey, D. Haussler, T. Hubbard, 
METHOD     and C. Chothia, http://cyrah.med.harvard.edu/~jong/assess_final.html, 1998. 
METHOD [5] S. Henikoff and J. C. Henikoff, JMB, vol 243, pp 574-578, Nov 1994. 
METHOD [6] K. Sjolander, K. Karplus, M. P. Brown, R. Hughey, A. Krogh, I. S. 
METHOD    Mian, and D. Haussler, CABIOS 12(4):327-345, 1996. 
METHOD  
METHOD  
METHOD Results 
METHOD  
METHOD Based on the hint that T0054 is "quite unique in structure", we 
METHOD expected to make a new-fold prediction unless we got a reasonably 
METHOD strong score. 
METHOD  
METHOD Our top hits for T0054 were homologs 1hmf, 1hma, 1aab and 1hryA; 
METHOD homologs 2minA and 1pedA; 1mioA and 1mioC; 1vhh; and 1prcC.  The top 
METHOD score was a modest -5.37, which is in a range where 80% of the hits 
METHOD are false positives.  Most of the better-scoring alignments aligned 
METHOD only small segments of T54, on the order of 20 residues. 
METHOD  
METHOD We see no reason to challenge the hint that T0054 is "quite unique in 
METHOD structure". 
MODEL  1 
PARENT NONE 
TER 
END 
