
PFRMAT AL
TARGET T0105
AUTHOR 1484-7979-5218
METHOD
METHOD  The results reported here are obtained using threading approach
METHOD  to the protein recognition problem.
METHOD
METHOD  There are two essential components to threading:
METHOD  (a) finding an optimal alignment (with gaps) of a sequence
METHOD      into a structure
METHOD  (b) scoring different alignments and deciding on the best
METHOD      matching shape.
METHOD
METHOD  To find an alignment we use standard dynamic programming
METHOD  algorithms.
METHOD
METHOD  To design score (energy) functions we use linear programming.
METHOD  In particular, a new model blending the higher prediction
METHOD  capacity of pairwise models with efficiency of profile
METHOD  potentials is optimized.
METHOD
METHOD  We call it THreading Onion Model 2 (THOM2), since it employs
METHOD  the first and the second contact shell to characterize the
METHOD  structural environment of a given site.
METHOD
METHOD  THOM2 model mimics effective pairwise interaction and
METHOD  incorporates significant cooperativity (N-body) effects.
METHOD  Linear programming is also used to determine optimal energy
METHOD  parameters for gaps. The new model provides an efficient and
METHOD  accurate threading approach, that can be used for genomics
METHOD  annotations.
METHOD
METHOD  The results are generated automatically (see LOOPP server
METHOD  http://ser-loopp.tc.cornell.edu/loopp.html) and include
METHOD  the best global and local threading (sequence to structure)
METHOD  alignments, as well as (structurally biased) local sequence
METHOD  to sequence alignments. The best matches according to a
METHOD  consensus score are reported as well.
METHOD
METHOD  First, matching of the whole query sequence(s) into whole
METHOD  library structures is evaluated using global variant of the
METHOD  dynamic programming algorithm and the novel THOM2 threading
METHOD  potential.
METHOD
METHOD  Next, the same THOM2 potential and a local variant of the
METHOD  dynamic programming algorithm are used to evaluate
METHOD  compatibility between best matching fragments of the query
METHOD  sequence(s) and library structures.
METHOD
METHOD  Finally, the best local sequence to sequence alignments
METHOD  are found using BLOSUM50 substitution matrix in conjunction
METHOD  with a gap penalty defined by a structural environment
METHOD  (number of neighbors) at a given site.
METHOD
METHOD  Our primary goal is to assign plausible structures to
METHOD  sequences that do not resemble significant sequence
METHOD  similarity to structurally (and functionally) chracterized
METHOD  proteins. To that end our primary predictions are based on
METHOD  threading and are complementary to standard sequence
METHOD  searches like BLAST or FASTA.
METHOD
METHOD  Accordingly, the contributions of the global and local
METHOD  threading alignments to the consensus scores outweigh
METHOD  contributions of the sequence alignments.
METHOD  However, especially when there is no sufficiently similar
METHOD  structure in the fold library, a marginal sequence
METHOD  similarity can be used to enhance the confidence
METHOD  of predictions.
METHOD
METHOD  For the CAFASP and CASP submissions the alignmnets for
METHOD  (up to) five best consensus matches are translated into
METHOD  an appropriate format, as included below. Since there
METHOD  may be three alignments contributing to the conseus score,
METHOD  the local threading alignments are chosen in the first
METHOD  place. If there is no contribution due to local threading
METHOD  alignment, then global threading alignmnet is used.
METHOD  If there is no contribution from threading, then local
METHOD  sequence to sequence alignment is submitted.
METHOD
METHOD  For details, please see http://www.tc.cornell.edu/CBIO/loopp/.
METHOD  Reference:  J. Meller and R. Elber, "The design of an
METHOD  efficient and accurate threading algorithm: Choice of energies
METHOD  and statistical verifications", Proteins, submitted.
METHOD
MODEL 3
REMARK Weak signal local sequence alignment
REMARK Local sequence alignment used as a model
PARENT 2reb
F  600      F  255
K  601      K  256
Q  602      Q  257
S  603      A  258
E  604      E  259
L  605      F  260
P  606      Q  261
V  607      I  262
T  608      L  263
C  609      Y  264
G  610      G  265
E  611      E  266
G  614      G  267
T  615      I  268
L  616      F  270
Y  617      Y  271
K  618      G  272
E  619      E  273
R  620      L  274
F  621      V  275
K  622      D  276
Q  623      L  277
G  624      G  278
T  625      V  279
S  626      K  280
K  627      E  281
K  628      K  282
C  629      L  283
I  630      I  284
Q  631      E  285
S  632      K  286
E  633      A  287
D  634      G  288
K  636      A  289
W  637      W  290
F  638      Y  291
T  639      S  292
P  640      Y  293
R  641      K  294
E  642      E  296
E  644      K  297
I  645      I  298
E  646      Q  300
G  647      G  301
D  648      K  302
R  649      A  303
G  650      N  304
A  651      A  305
S  652      T  306
K  653      A  307
W  655      W  308
TER
END





