POPULATE USER PROJECT DIRECTORY WITH INPUT FILES

In this project, we attempt structure prediction for protein G, a
small 56 residue protein.  A high-resolution crystal structure for
protein G is available in PDB entry 1pga.  In preparation for the
calculations, the "/up0" directory tree is populated with 3 input
files, "/up0/exp/1pga.pdb", "/up0/exp/seq.1pga", and
"/up0/stp/stp.full.1pga", where "1pga.pdb" is the crystal structure,
"seq.1pga" is the residue sequence in 1-letter code format, and
"stp.full.1pga" specifies a subset of degrees of freedom that includes
all torsion angles.

CALCULATE FULL ENERGY OF THE NATIVE CONFORMATION

We begin by establishing the full energy of the native conformation,
where the native conformation is taken to be the conformation
determined experimentally for the crystal.

% ereg up0 1pga

directs local energy minimization starting from the crystal structure.
Commands are executed from directory "/src/str".

Summary of the "ereg" command.
|SYNTAX
|  ereg up0 MOL
|INPUT FILES                   OUTPUT FILES
|/up0/exp/MOL.pdb              /up0/dgn/ereg.MOL
|                              /up0/seq/seq.MOL
|                              /up0/tor/tor.MOL.CNF
|                              /up0/car/MOL.CNF.pdb

From file "up0/dgn/ereg.1pga", the full energy of the energy minimized
crystal structure, "up0/car/1pga.06.pdb", is -1630.58 kcal/mol.

% rcyc up0 1pga 06

directs energy refinement starting from the endpoint of local energy
minimization.

Summary of the "rcyc" command.
|SYNTAX
|  rcyc up0 MOL CNF
|INPUT FILES                   OUTPUT FILES
|/up0/seq/seq.MOL              /up0/dgn/rcyc.MOL
|/up0/tor/tor.MOL.CNF          /up0/stp/stp.g???.MOL
|                              /up0/car/MOL.g???.pdb
|                              /up0/tor/tor.MOL.g???

% cp ../../up0/car/1pga.g013.pdb ../../up0/exp/1pga_r.pdb
% ereg up0 1pga_r

copies and renames the endpoint of energy refinement, and directs a
second local energy minimzation starting from this endpoint.  From
file "up0/dgn/ereg.1pga_r", the full energy of the energy refined
native conformation, "up0/car/1pga_r.05.pdb", is -1663.08 kcal/mol.

% cp ../../up0/tor/tor.1pga_r.05 ../../up0/tor/tor.1pga.n000
% cp ../../up0/car/1pga_r.05.pdb ../../up0/car/1pga.n000.pdb

assigns a more meaningful name for the energy refined native
conformation.
 
TEST STABILITY OF THE NATIVE CONFORMATION

As a test of the stability of the native conformation

% ptra up0 1pga n000 full

directs a trajectory search starting from the refined native
conformation.

Summary of the "ptra" command.
|SYNTAX
|  ptra up0 MOL CNF SUB
|INPUT FILES                   OUTPUT FILES
|/up0/seq/seq.MOL              /up0/dgn/ptra.MOL.CNF.SUB
|/up0/tor/tor.MOL.CNF          /up0/car/MOL.t???.pdb
|/up0/stp/stp.SUB.MOL          /up0/tor/tor.MOL.t???

Parameter NCYCLES in file "/src/str/thread_config" controls the number
of cycles from which the trajectory search is formed.  For this
project, NCYCLES is set to 128.

From file "/up0/dgn/ptra.1pga.n000.full", the final section headed
"COMPACT SUMMARY" shows that no conformation is found such that full
energy is reduced relative to the starting conformation "n000".  This
result, the inability of trajectory search to unfold the refined
native conformation, validates the current energy function.

GENERATION OF IGOR MODEL FOLDS

Structure prediction begins with generation of a collection of 16 igor
model folds.

% igor up0 1pga

Summary of the "igor" command.
|SYNTAX
|  igor up0 1pga
|INPUT FILES                   OUTPUT FILES
|/up0/exp/seq.1pga             /up0/dgn/igor.1pga
|                              /up0/dgn/igodmp.1pga
|                              /up0/seq/seq.1pga_??i
|                              /up0/tor/tor.1pga_??i.igo
|                              /up0/car/1pga_??i.igo.pdb

For each fold 1pga_??i, where ?? are 2 digits ranging from 00 to 15,
energy refinement creates a starting conformation for global energy
minimization using trajectory search.

% rcyc up0 1pga_??i igo

The endpoints of energy refinement are renamed using a more meaningful
conformation name "s000".

% cp ../../up0/tor/tor.1pga_??i.g013 ../../up0/tor/tor.1pga_??i.s000
% cp ../../up0/car/1pga_??i.g013.pdb ../../up0/car/1pga_??i.s000.pdb

In preparation for trajectory searches, "stp" files directing motion
with respect to the full set of generalized degrees of freedom are
copied from the previously created file "stp.full.1pga".

% cp ../../up0/stp/stp.full.1pga ../../up0/stp/stp.full.1pga_??i

TRAJECTORY SEARCHES FROM IGOR MODEL FOLD STARTING POINTS

For each fold 1pga_??i, a trajectory is generated in segments
consisting of 128 cycles.

% ptra up0 1pga_??i s000 full

A trajectory is discontinued if, following generation of a 128 step
segment, either 1) energy remains high relative to energies for
parallel trajectories starting from other igor folds, or 2) the 128
cycle segment fails to lower energy relative to the previous segment.

Energy (kcal/mol) as a function of cycle number for trajectory
searches starting from igor folds.
____________________________________________________________
fold       cycle number
         0    128    256    384    512
____ ______ ______ ______ ______ ______
  00  -1480  -1562  -1562
  01  -1538  -1538
  02  -1528  -1598  -1604  -1602  -1592
  03  -1503  -1583  -1583
  04  -1472  -1492
  05  -1532  -1563  -1563
  06  -1448  -1566  -1566
  07  -1492  -1508
  08  -1497  -1590  -1591  -1607
  09  -1441  -1488
  10  -1468  -1569  -1603  -1603
  11  -1471  -1545  -1548
  12  -1497  -1556  -1560  -1578
  13  -1461  -1542  -1566  -1566
  14  -1511  -1531
  15  -1468  -1529
____________________________________________________________

Because of the large size of the output files created by the "ptra"
command, files "/up0/car/MOL.t???.pdb" and "/up0/tor/tor.MOL.t???"
were removed, and files "/up0/dgn/ptra.MOL.CNF.SUB" were edited to
retain only the final compact summary of the trajectory.

CONCLUSIONS

The trajectory searches find local minima having lower energies
relative to the starting igor model fold conformations.

For igor model folds 02 and 10, the 2 starting points from which
search trajectories reach the lowest energies, the folds match the
crystal structure in predicted secondary structure elements and
packing topology.  For both of these folds, trajectory search
lengthens the alpha-helix, extends the beta-sheet, recovers the native
twist of the sheet, and reproduces the native geometry of packing
between the helix and sheet.

No conformation is found having a lower energy relative to the refined
native conformation.  The existence of such a conformation would
indicate some significant error remaining in the current energy
function.

Trajectories consisting of 512 or fewer cycles are unable to recover
the refined native conformation.  Either longer search trajectories
using the current algorithm or a more efficient algorithm would be
required to achieve more effective global energy minimization for
systems consisting of small proteins.
