
PFRMAT AL
TARGET T0110
AUTHOR 3670-4530-6947
METHOD Overview
METHOD 
METHOD Fold recognition for this target was performed using the SAM-T99 and
METHOD SAM-T2K methods (which are similar to SAM_T98 [3]) using SAM version
METHOD 3.1 [1], a refinement of the methods developed by this group for CASP3
METHOD [7]).  These methods attempt to find and multiply align a set of
METHOD homologs to a given sequence, then create an HMM from that multiple
METHOD 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 publications [3,4], uses Dirichlet
METHOD mixtures [6] to regularize the counts and an entropy method to set the
METHOD final weights.
METHOD 
METHOD We are currently using SAM-T2K to generate the HMM from the target
METHOD sequence, but are still using the library of SAM-T99 template HMMs,
METHOD since the SAM-T2K method is not yet fully stable.
METHOD 
METHOD Alignment generation
METHOD 
METHOD The initial step uses WU-BLASTP to search NRP to get nested sets of
METHOD possible homologs---from a set of very similar sequences to a set of
METHOD possibly related sequences.
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 most similar sequences found by BLASTP are
METHOD used as the search set.  The threshold is set very strictly, so that
METHOD only good matches to the sequence are considered.  
METHOD 
METHOD On subsequent iterations the input alignment is the output from the
METHOD previous iteration, the search set is a larger set of possible
METHOD homologs found by BLASTP, and the thresholds are gradually loosened.
METHOD 
METHOD The HMM used for scoring and aligning sequences is built from a
METHOD multiple alignment using the w0.5 script, which aims to get an average
METHOD of 0.5 bits of information per column of the alignment.
METHOD 
METHOD All PDB protein sequences (including, unfortunately, theoretical
METHOD models) are scored with the HMM built from the target alignment, and
METHOD all the template HMMs in our SAM-T99 library (about 3760) are used to
METHOD score the target sequence.
METHOD 
METHOD High-scoring hits in either direction are examined by hand, as are
METHOD other potential targets found by public servers (see the CAFASP
METHOD experiment) or by functional considerations.
METHOD 
METHOD The final alignment is selected from among the various alignments
METHOD obtained by varying whether the template model or the target model is
METHOD used for the alignment, whether local or global alignment is chosen,
METHOD and various other parameters.  In some cases parts of different
METHOD alignments are combined.
METHOD 
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 [7] Karplus, K; Barrett, C; Cline, M; Diekhans, M; Grate, L; Hughey, R. 
METHOD     Predicting protein structure using only sequence information.
METHOD     Proteins, 1999, Suppl 3:121-5.
METHOD 
METHOD 
METHOD We got a good multiple alignment from SAM-T2k, which gave us a clean
METHOD secondary-structure prediction.
METHOD 
METHOD Unfortunately, we were unable to find any existing structures with
METHOD strong matches to T0110, with either the template or the target HMMs.
METHOD We got some weak predictions for matches to 1egaA and 1lehA using
METHOD 2-track HMMs (amino acid and predicted secondary structure), both of
METHOD which had flat 3-strand sheets with a helix between the second and
METHOD third strands.  For 1egaA we also matched the first helix, and for
METHOD 1lehA we matched the final helix.
METHOD 
METHOD We combined these predictions using our untested mini-threader
METHOD "undertaker".
METHOD 
METHOD The resulting prediction is not really as compact as we'd like, as the
METHOD final helix does not pack well against the sheet, probably because the
METHOD turn between the first and second strands is bent the wrong way and is
METHOD interfering with the packing of the final helix.  Undertaker does not
METHOD currently have much chance of flexing the turn out of the way without
METHOD destroying the beta sheet.
METHOD 
MODEL 1
PARENT 1ega_A
R 7 Q 190
S 8 R 191
D 9 F 192
R 10 M 193
V 11 A 194
A 12 S 195
Q 13 E 196
E 14 I 197
I 15 I 198
Q 16 R 199
K 17 E 200
E 18 K 201
I 19 L 202
A 20 M 203
V 21 R 204
I 22 F 205
L 23 L 206
Q 24 G 207
R 25 A 208
E 26 E 209
V 27 L 210
P 30 P 211
R 31 Y 212
I 32 S 213
G 33 V 214
M 34 T 215
V 35 V 216
T 36 E 217
V 37 I 218
S 38 E 219
D 39 R 220
V 40 F 221
E 41 V 222
V 42 S 223
S 43 N 224
S 44 E 225
D 45 R 226
L 46 G 227
S 47 G 228
Y 48 Y 229
A 49 D 230
K 50 I 231
I 51 N 232
F 52 G 233
V 53 L 234
T 54 I 235
F 55 L 236
L 56 V 237
F 57 E 238
D 58 R 239
E 61 E 240
M 62 G 241
A 63 Q 242
I 64 K 243
E 65 K 244
Q 66 M 245
G 67 V 246
M 68 I 247
K 69 G 248
G 70 N 249
L 71 K 250
E 72 G 251
K 73 A 252
A 74 K 253
S 75 I 254
P 76 K 255
Y 77 T 256
I 78 I 257
R 79 G 258
S 80 I 259
L 81 E 260
L 82 A 261
G 83 R 262
K 84 K 263
A 85 D 264
M 86 M 265
R 87 Q 266
L 88 E 267
R 89 M 268
I 90 F 269
V 91 E 270
P 92 A 271
E 93 P 272
I 94 V 273
R 95 H 274
F 96 L 275
I 97 E 276
Y 98 L 277
TER
END

