|
Jul 1 2004
|
[ addBfactor | alignMinCoverage | alignOldStatWeight | axisLength | clashThreshold | cnWeight | consensusStrength | dcWeight | CONSENSUS_strength | densityCutoff | dielConst | dielConstExtern | drop | fogStart | gapExtension | gapOpen | hbCutoff | lineWidth | mapSigmaLevel | mcBell | mcJump | mcShake | mcStep | mfWeight | mimelDepth | mimelMolDensity | r_out | r_2out | resLabelShift | rsWeight | selectMinGrad | selectSphereRadius | sfWeight | shininess | ssThreshold | ssWeight | ssearchStep | surfaceTension | tempLocal | temperature | timeLimit | tolGrad | tzWeight | vicinity | vwCutoff | vwExpand | vwSoftMaxEnergy | waterRadius | wireBondSeparation | xrWeight ]
ICM-shell real variables are the following.
additional B-factor which may be added to the current atomic B-values
to create a smoother electron density map from a set of atoms.
See also:
make map
Default (0.0)
a threshold for the ratio of the aligned residues
to the shorter sequence length. All alignments shorter than
alignMinCoverage*minLength will not be reported by
find database command.
The default value is 0.5. However the parameter can
be tuned with the respect to the database and the nature of the
query sequence.
- Search against the protein domain sequence database: use 0.5 or higher
- Search a multidomain sequence against long multidomain sequences:
use 0.1 or lower
See also: 'alignMinMethod , find database .
Default (0.5)
a parameter influencing the statistical evaluation of sequence comparison significance
in the find database command.
Statistical significance can be evaluated in two ways: first, a priori, i.e. before the
database search and based only on the individual score of an alignment of interest and its
theoretical distribution, or, second, a posteriori, i.e. on the fly and on the basis of
all empirically observed scores of other alignments in the course of the database search.
The parameter ranges from 0. to 1. and sets how two different
statistical criteria of alignment significance, a precomputed (the old one)
and a run-time, should be mixed.
Zero corresponds to only the run-time measure ( the new way) in which the
significance is evaluated on the run-time statistics of the observed alignment scores,
while one corresponds to the statistics evaluated before the search using the
formula from Abagyan and Batalov, 1997 .
If the database is small then the run-time score statistics may be incomplete
and alignOldStatWeight closer to 1. is a better choice. On the other hand,
the run-time statistics has several principal advantages:
| Precomputed statistics (1.) based on individual alignment score and length | Run-time statistics (0.) based on distribution of scores |
|---|
| works always | relies on database diversity |
| is trained only in 64 condition sets and ZEGA alignment | automatically adjusts to any set of conditions, e.g. gapFunction, or alignMinCoverage |
| does not reflect compositional bias | automatically reflects all seq. properties |
| does not reflect extra terms to the score | accounts for solvent accessibility correction (see accFunction ) |
The run-time statistics will fit the scores to an optimized
empirical function. This function avoids the problems of the normal
distribution, and certain pitfalls of a popular EVD function.
The resulting P-value is a reliable estimate of the false positive
rate if the database is sufficiently diverse, i.e. the fraction
of sequences similar to the query is small.
For example, searching a tyrosine kinase through a database of
tyrosine kinases will yield incorrectly low pP-values (pP = -Log(P)).
Reliable expect-values: P * Nof(sequences) <= 0.1 .
Example:
- Swissprot has N=89,000 sequences. LogN = 4.95
- Reliable pP = LogN + 3, twilight pP is from LogN + 1. to LogN + 3.
length (in Angstroms) of the X,Y,Z axes of the coordinate frame. The axes
can be displayed by the display origin or display virtual command.
The axes are marked X Y Z . Example:
buildpep "ala ala his his"
display
axisLength=10.
display origin
Default (1.5)
a clash is defined as an interatomic distance less than a sum of van
der Waals radii of two atoms of interest multiplied by the clashThreshold
parameter. For hydrogen bonded atoms, the distance threshold is additionally
reduced by 20% .
See also: display clash , show clash, GRAPHICS.clashWidth
Default (0.82)
weighting factor for the interatomic
distance restraints
penalty term.
See also: tzMethod , drestraint and Bfactor .
Default (1.0)
a real parameter between 0. and 1. controlling
the percentage of sequence identity in an alignment column
required to establish a consensus. This parameter r is
applied to the C0 parameters defined in the CONSENSUS table
according to this formula:
C = C0 + (100-C0)*2*(r-0.5)
The 0.5 value corresponds to the percentage in the table.
Default (0.5)
weighting factor for the density correlation term
"dc".
Default (1.0)
regulates the strength of consensus modifying the CONSENSUS.fraction values.
The CONSENSUS table controls the rules of consensus derivation from
an alignment . This table may look like this:
#>T CONSENSUS
#>-symbol------fraction----residues---
A 80 A
C 90 C
D 85 D
d 60 ND
...
...
The CONSENSUS_strength (denoted S ) parameter can increase or decrease the fraction
values f according to the following formula:
f1 = f + (100-f)*(S-0.5)
Therefore if S = 1. all fraction values become equal to 100% .
This affects the Consensus function and the GUI representation of
alignment consensus in ICM versions above 3.0.
To color structures according to the consensus, use the color as_ alignment
command, or, interactively, left-click on a color icon and select
ColorBy followed by alignment.
Default (0.5)
The neglected fraction of the total atomic electron density in the course of
calculation of the grid electron density from atomic positions.
Atomic density distribution is approximated by two gaussian
functions which need to be truncated for computational efficiency.
See also:
make map command and related operations with the electron density,
penalty term "dc"
Default (0.1)
dielectric constant of the solute used in Coulomb, distance-dependent,
MIMEL, and boundary element electrostatic calculations.
If electroMethod="distance dependent" the actual dielectric constant
is a product of dielConst and a distance from a change.
See also: dielConstExtern, term "el"
Default (4.0)
dielectric constant of the solvent exterior used in MIMEL and boundary
element
electrostatic
calculations.
Default (78.5)
expected initial function drop in local
minimization.
The parameter is used to evaluate initial step size. If your function is
already very close to its minimum, it is a good idea to reduce the
parameter, otherwise the procedure will start with an inappropriately large step.
Default (10.0)
relative Z-depth with respect to the front clipping plane at which fogging starts.
With this parameter you can keep some area in front without any fog and than
start gradually increasing the effect until the back clipping plane.
To activate fog use Ctrl-D .
Clipping planes can be moved with Ctrl-MiddleMB (front plane)
and MiddleMB - left 5% margin (back plane).
Actually the mapping of these operations to particular keystrokes
is flexible and is defined in the icm.clr file. For Linux it is
useful to redefine the back-clipping plane movements to
mode 9 Right5-Mid # Move rear clipping plane
Right5 means that you use the 5% right margin of your window.
Usually the fog color is the same as the background color. You can
change the fog color with the
color volume Color
command.
Default (0.3)
Relative gap extension penalty used in an
alignment
procedure. The absolute gap penalty is calculated as a product of
gapExtension
and the average diagonal element of the
residue comparison table
Default (0.15)
See also
gapFunction , Align.
Relative gap opening penalty used in an alignment
procedure. The absolute gap penalty is calculated as a product of
gapOpen and the average diagonal element of the
residue comparison table
You may vary gapOpen between 1.8 and 2.8 to analyze dependence of your
alignment on this parameter. Lower pairwise similarity may require somewhat
lower gapOpen parameter. A value of 2.4 (gapExtension=0.15) was shown to be
optimal for structural similarity recognition with the
Gonnet et. al.)
matrix, while a value of 2.0 was optimal for the
Blosum50)
matrix (
Abagyan and Batalov, 1997).
Default (2.4).
See also
gapFunction , Align.
(Angstroms) cutoff radius for
hydrogen bonding
interactions.
Default (3.0)
the real width of lines used to display the wire representation of chemical bonds.
See also IMAGE.lineWidth parameter which controls line thickness in molecular
images generated by the write postscript
command, and the PLOT.lineWidth which controls the
width for the plotcommand.
Default (1.0)
Example:
build string "se nad" # NAD molecule
lineWidth = 3.
wireStyle="chemistry"
display
(in Rmsd values over the mean value). Margin value used for making
graphical objects contouring the 3D density map .
Default (1.5)
average relative size of normally distributed
montecarlo
step from the center of an ellipsoid surrounding the multidimensional
variable restraint
zone.
Example:
mcBell = 1.0 # places one standard deviation at the rs border
mcBell = 2.0 # distribution is two times broader etc.
Default (1.0)
maximum value (in degrees) of random angular distortion per variable. This
local random perturbation occurs if visitsAction,
highEnergy or rejectAction
ICM-shell variables are set to the "`random" value.
Randomization is a possible action in three problematic situations in
montecarlo procedure.
Default (30.0)
amplitude [Angstrom] of Brownian type the montecarlo
random move applied to a molecule when one of the 6 variables defining its
relative position is picked. Usually these variables may be selected
by v_myMolecule//?vt*selection. The center of mass of the molecule
randomly moves in an xyz sphere of mcShake radius.
The molecule is also randomly rotated around a random axis
with an amplitude equal to mcShake divided by the MolecularRadius .
This parameter is also used as a default amplitude for the randomize command
where the six position/orientation variables are selected.
Default (2.0)
montecarlo step size (degrees).
Maximum random change of one variable. This parameter
is also used as the default amplitude for the
randomize
command
Default (180.0)
the overall weighting factor for the "mf" penalty term.
This term may contain any user-defined
energy or penalty function depending on pairs of atom types and
interatomic distances.
The parameters for the term are stored in icm.pmffile.
The weighting factor will determine the "mf" term contribution
with respect to the energy terms.
See also: "mf" term,
cnWeight, dcWeight, rsWeight, ssWeight, tzWeight, ssWeight .
Default (1.0)
The fraction of an estimated molecular radius which is taken as a radius
of the probe sphere used by the MIMEL
algorithm. The accessible surface of this probe sphere is used to calculate
the distance between a charge and the effective dielectric boundary.
Described in detail on p. 991-992 of (`ato94{ Abagyan and Totrov, 1994}). For small molecules
mimelDepth = 0.3 is recommended.
Default (0.5).
a coefficient used to calculate the effective molecular radius
from a number of atoms. Recommendation: do not touch it, unless you
are an advanced user. See also the description of the
MIMEL method.
Default (1.0).
a real variable where some commands and functions (e.g.
show area,
show volume,
superimpose,
minimize tether,
Corr , Axis , Align) store their output.
Also, in the electrostatic calculations with the
MIMEL or REBEL method, the solvation
energy part of the electrostatic energy is returned in r_out.
Default (0.0). See also:
r_2out.
a real variable where some commands and functions
(e.g. Axis ) store their output.
Default (0.0). See also: r_out .
is the translation towards the viewer (normal to the graphics screen)
used to display a label in front of cpk's or skin's
rather than bury the label under them.
The recommended value is 4.
See also:
resLabelStyle
Default (0.0) to be used with more popular wire representation.
weighting factor for the multidimensional variable restraints
penalty term.
Default (1.0).
default minimal gradient vector length for gradient atom selection (
a_//G). This parameter is also used by the
montecarlo fast command, which
requires a value of 2. to 10. for optimal performance.
Example:
read pdb "1fox"
convertObject a_ yes no yes no
show energy
selectMinGrad=80.
show a_//G
display
display a_//G cpk
Default (1.5).
default sphere radius (in Angstroms) for atom selections in Sphere( ) function,
as well as the gaussian 3D averaging radius in the
color ribbon command with ribbonColorStyle="reliability".
This parameter is also used in the compare surface command.
Default (5.0).
the overall weighting factor for the surface solvation energy term.
If surfaceMethod is "constant tension" it can also be controlled by
the surfaceTension parameter.
Default (1.0).
parameter defining the shininess of solid surfaces such as
cpk, ribbon, ball, stick, xstick, and skin when they are
displayed.
Only values in the range [0.,128.] are accepted.
Example in which we generate a high quality CPK image:
buildpep "ASDW"
GRAPHICS.quality = 15.
shininess = 100.
display cpk
Default (20.0). Range: from 0. to 128.
threshold distance between two Sg atoms of cystein residues.
This distance controls the automatic formation of disulfide
bonds in some commands (e.g.
read pdb).
Default (2.35).
weighting factor for the disulfide bridge
( "ss"
) penalty term.
Default (1.0).
angular increment (in degrees) for variables in the systematic search
( ssearch command ) in so called "local"
mode when the search is performed around the current conformation.
Default (30.0).
surface energy density in kcal/mole/A2. The surface energy which
is a product of this parameter by the total solvent accessible area
will be stored in the "sf" term, if
surfaceMethod preference is set to "constant tension" .
Note, that if a part of the system is represented with grid potentials,
one needs a special m_ga grid map for correct calculations of the
surface accessibilities.
Default (0.012)
montecarlo
simulation temperature for local deformation random moves.
This temperature can be set higher than the normal
temperature since a local deformation
includes a larger number of variables and may require a higher
temperature for efficient sampling. To set the same simulation temperature,
specify:
tempLocal=temperature
in your script.
Default (5000.).
montecarlo
simulation temperature. A new trial conformation with a higher energy
than the current one is accepted with the probability of
exp(-(Etrial - Enew)/RT)). RT is 0.6 kcal/mole for T = 300 Kelvin.
The effect of temperature on the montecarlo procedure is the following:
- to find the global minimum successfully one needs a combination of persistense
if a chosen place with a good sense of when to stop searching in this place and
move along to the next one.
- if the temperature is too high, the acceptance ratio improves (gets higher)
and wider sampling becomes easier since more high energy conformations are
accepted. The downside of this is the low "persistence" (or "lack of patience")
of the search procedure. Instead of spending more time in each conformational
vicinity to find the real global minimum,
the procedure just tries a couple of sub-optimal conformations and jumps away.
- if the temperature is too low the procedure may not cover the global conformational
space of interest.
Default (300.).
the real running time of ICM in hours (wall time) before a long simulation is interrupted.
This limit can interrupt a montecarlo procedure (another exit condition is mncallsMC),
Example:
timeLimit = 2. # two hours
montecarlo # will interrupt the simulation in 2 hours
Default (99999.0).
gradient tolerance criterion for local minimization.
Minimization is stopped if the gradient root-mean-square deviation
from zero is less than the parameter value.
Default (0.05).
the overall weighting factor for the tether penalty term.
You may need to increase it while minimizing a highly
energetically strained molecule resulting from the initial steps of the
conversion or regularization
procedure.
Additional atom specific weights can be introduced through atomic bfactors with
tzMethod="weighted"
Default (1.0).
maximum angular root-mean-square deviation per variable (degrees) or
cartesian root-mean-square deviation per atom (Angstroms)
when two structures are still considered belonging to the same
conformational family in conformational stack manipulations.
The type of comparison is defined by the compare command.
Examples:
compare a_//ca,c,n # compare by Cartesian RMSD
vicinity = 3.0 # conf. are similar if RMSD< 3 A
compare v_//phi,psi # compare by angular RMSD
vicinity = 40.0 # conf. are similar if aRMSD < 40 deg
Default (15.0) . Do not forget to set it to
a lower value if Cartesian RMSD is
compared.
(Angstroms) cutoff radius for van der Waals interactions and
Coulomb electrostatics .
Default (7.5).
radius of a probe sphere used to
display
a dotted
surface
of a molecule. All van der Waals radii are expanded by this value.
vwExpand=0 corresponds to the
CPK
surface, vwExpand=1.4 corresponds to the water-accessible surface.
Be aware of the difference between the
waterRadius
and
vwExpand parameters:
waterRadius is used in
- show energy "sf"
- show [area|volume] skin
- display skin
while vwExpand is used in
- show [area|volume] surface
- display surface
Default (1.4).
Parameter defining maximal energy value of van der Waals repulsion at r -> 0. for
the finite approximation van der Waals function ( vwMethod = "soft" ).
This parameter must be greater than 0. kcal/mole.
Note that in the "soft" mode, the electrostatic energy will be automatically
buffered to avoid singularities. You will see that the electrostatic term "el"
changes upon switching from vwMethod=1 to vwMethod=2 .
Default (7.0).
radius of water sphere which is used to calculate
an analytical molecular surface (referred to as skin) as well as
the solvent-accessible surface (centers of water spheres).
Because of the complexity of skin calculations, it is not recommended that one play's with this
parameter (of course, you rushed to do exactly that). Be aware of the difference
between the waterRadius and vwExpand parameters: waterRadius is used in
- show energy "sf"
- show [area|volume] skin
- display skin
while vwExpand is used in
- display surface
- show [area|volume] surface
Default (1.4).
the distance between two parallel lines representing a chemical double bond if
wireStyle = "chemistry".
Default ( 0.15 Angstroms).
the overall weighting factor for the
structure factor correlation
penalty term.
See also: xrMethod .
Default (1.0).
|