manual.f.html

最经典的分子对结软件

</TD>
</TR>
<TR>
<TD ROWSPAN="1" COLSPAN="1">
<P CLASS="Normal">
<A NAME="pgfId=51428">
 </A>
ext</P>
</TD>
<TD ROWSPAN="1" COLSPAN="1">
<P CLASS="Normal">
<A NAME="pgfId=51430">
 </A>
File extension (*.mol2, *.pdb, etc.).</P>
</TD>
</TR>
</TABLE>
<P CLASS="Body">
<A NAME="pgfId=63642">
 </A>
If <A HREF="Manual.19.html#29408" CLASS="XRef">
torsion_drive</A>
 has been selected, then the torsion positions of the intervening bond are searched when each segment is reattached.  If <A HREF="Manual.19.html#32915" CLASS="XRef">
torsion_minimize</A>
 has been selected, then the intervening torsion may be relaxed.  Minimization of the bond is performed in isolation, or in concert with inner torsions if the <A HREF="Manual.19.html#19627" CLASS="XRef">
reminimize_layer_number</A>
 parameter is set to a non-zero value.  Relaxing multiple layers helps prevent the search from getting stuck in dead-ends.  Although computationally expensive, the position of the anchor may be simultaneously optimized during the conformation search with the <A HREF="Manual.19.html#29269" CLASS="XRef">
reminimize_anchor</A>
 parameter.  When all segments have been added, the entire molecule may be relaxed if the <A HREF="Manual.19.html#40569" CLASS="XRef">
reminimize_ligand</A>
 parameter is set.</P>
<DIV>
<H6 CLASS="Heading-5">
<A NAME="pgfId=10111">
 </A>
Docking with the Anchor-First procedure</H6>
<P CLASS="Body">
<A NAME="pgfId=46512">
 </A>
The process of docking a molecule using the anchor-first strategy is shown in <A HREF="Dock.Manual.3.html#15583" CLASS="XRef">
Figure 5</A>
.  The  amount of searching is under full user control.  The process begins with docking the anchor.  This step is controlled with the <A HREF="Manual.19.html#34562" CLASS="XRef">
orient_ligand</A>
 parameters (please refer to <A HREF="Manual.e.html#30227" CLASS="XRef">
Orientation Search on page 28</A>
) and results in No anchor positions.  The conformation search begins by pruning these orientations according to rank and position (see <A HREF="Manual.f.html#30944" CLASS="XRef">
Pruning the conformation search tree</A>
 below) to produce Nc positions.  Each subsequent cycle of the conformation search involves expanding the ensemble of partially built binding positions by adding a new segment and performing a torsion search on the newly formed bond and then contracting the ensemble with pruning.  The torsion search on each newly formed bond results in an expansion of the set of Nc partial configurations to NcNt configurations.  The size of Nt  is based on the number of increments used for the current bond and can be modified by altering the entries in the <A HREF="Manual.4c.html#25230" CLASS="XRef">
flex_drive.tbl</A>
 file.  The expanded set of binding positions is then pruned back to Nc configurations.  The conformation search continues expanding and pruning the set of partial binding position until each binding position represents a complete molecule.</P>
<P CLASS="Body">
<A NAME="pgfId=46259">
 </A>
This search technique is particularly useful for docking, but it also may be used for conformation analysis and stand-alone minimization.</P>
</DIV>
<DIV>
<H6 CLASS="Heading-5">
<A NAME="pgfId=46099">
 </A>
<A NAME="30944">
 </A>
Pruning the conformation search tree</H6>
<P CLASS="Body">
<A NAME="pgfId=45149">
 </A>
During each cycle of the conformation search, the expanded set of partial configurations is pruned based on the setting of <A HREF="Manual.19.html#12399" CLASS="XRef">
configurations_per_cycle</A>
.  The pruning attempts to retain the best, most diverse configurations using a top-first pruning method which proceeds as follows.  The configurations are ranked according to score.  The top-ranked configuration is set aside and used as a reference configuration for the first round of pruning.  All remaining configurations are considered candidates for removal.  A weighted root-mean-squared distance (wRMSD) between each candidate and the reference configuration is computed according to <A HREF="Manual.f.html#19488" CLASS="XRef">
Equation 1</A>
.</P>
<DIV>
<H6 CLASS="Equation">
<A NAME="pgfId=46368">
 </A>
<A NAME="19488">
 </A>
<IMG SRC="Manual-8.gif" ALIGN="BASELINE">
Equation 1</H6>
<P CLASS="Body">
<A NAME="pgfId=46360">
 </A>
where Li is the layer to which atom i is assigned.  The RMSD is weighted in this fashion to make it more sensitive to the position of the outer segments.  The outer segments are more important because they have a greater influence over the position of subsequently added segments.</P>
<P CLASS="Body">
<A NAME="pgfId=46382">
 </A>
Each candidate is then evaluated  for removal based on its rank and  wRMSD using the inequality shown in <A HREF="Manual.f.html#22575" CLASS="XRef">
Equation 2</A>
.  If the factor is greater than <A HREF="Manual.19.html#12399" CLASS="XRef">
configurations_per_cycle</A>
, the candidate is removed.  Based on this factor, a configuration with rank 2 and 0.2 Angstroms wRMSD is comparable to a configuration with rank 20 and 2.0 Angstroms wRMSD.  The next best scoring configuration which survives the first pass of removal is then set aside and used as a reference configuration for the second round of pruning, and so on.</P>
</DIV>
<DIV>
<H6 CLASS="Equation">
<A NAME="pgfId=45185">
 </A>
<A NAME="22575">
 </A>
<IMG SRC="Manual-9.gif" ALIGN="BASELINE">
Equation 2</H6>
<P CLASS="Body">
<A NAME="pgfId=45171">
 </A>
This pruning method attempts to balance the twin goals of recovering the best scoring and the most different binding configurations without introducing additional user parameters.  The pruning method replaces the hierarchical clustering method used in the initial release of dock version 4.0, because it is faster (N logN versus N2) and biases its search time towards molecules which sample a more diverse set of binding modes.  As the value of <A HREF="Manual.19.html#12399" CLASS="XRef">
configurations_per_cycle</A>
 is increased, the anchor-first method approaches an exhaustive search.</P>
</DIV>
</DIV>
<DIV>
<H6 CLASS="Heading-5">
<A NAME="pgfId=46245">
 </A>
Time requirements</H6>
<P CLASS="Body">
<A NAME="pgfId=46246">
 </A>
The time demand grows linearly with <A HREF="Manual.19.html#12399" CLASS="XRef">
configurations_per_cycle</A>
, the number of flexible bonds and the number of torsion positions per bond, as well as the number of anchor segments explored for a given molecule.  Using the notation in <A HREF="Dock.Manual.3.html#15583" CLASS="XRef">
Figure 5</A>
, the time demand can be expressed as</P>
<DIV>
<H6 CLASS="Equation">
<A NAME="pgfId=46330">
 </A>
<IMG SRC="Manual-10.gif" ALIGN="BASELINE">
Equation 3</H6>
<P CLASS="Body">
<A NAME="pgfId=46269">
 </A>
where the additional terms are:</P>
<TABLE>
<TR>
<TD ROWSPAN="1" COLSPAN="1">
<P CLASS="Normal">
<A NAME="pgfId=46291">
 </A>
Na</P>
</TD>
<TD ROWSPAN="1" COLSPAN="1">
<P CLASS="Normal">
<A NAME="pgfId=46296">
 </A>
is the number of anchor segments tried per molecule.</P>
</TD>
</TR>
<TR>
<TD ROWSPAN="1" COLSPAN="1">
<P CLASS="Normal">
<A NAME="pgfId=46298">
 </A>
Nb</P>
</TD>
<TD ROWSPAN="1" COLSPAN="1">
<P CLASS="Normal">
<A NAME="pgfId=46306">
 </A>
is the number of rotatable bonds per molecule.</P>
</TD>
</TR>
</TABLE>
</DIV>
</DIV>
<DIV>
<H6 CLASS="Heading-5">
<A NAME="pgfId=46271">
 </A>
<A NAME="38957">
 </A>
Manual specification of anchor segment</H6>
<P CLASS="Body">
<A NAME="pgfId=45087">
 </A>
The user can override the automatic anchor selection performed by dock by specifying a STATIC ATOM SET named ANCHOR in the molecule input file.  For an example, please see <A HREF="Manual.41.html#19711" CLASS="XRef">
SYBYL MOL2 format on page 99</A>
.  See the previous discussion of <A HREF="Manual.f.html#21011" CLASS="XRef">
Manual specification of non-rotatable bonds</A>
 for related instructions.  It must be pointed out that the user can include as many atoms as desired in the ANCHOR set, but only the first atom will be used.  The anchor segment which includes the ANCHOR atom will then be used as the segment anchor.  In order to make a larger anchor than would be produced using the automatic segmentation based on the location of rotatable bonds, the user will need to manually specify the necessary bonds as non-rotatable (with a RIGID BOND set).</P>
</DIV>
</DIV>
<DIV>
<H3 CLASS="Heading-4">
<A NAME="pgfId=45093">
 </A>
<A NAME="15938">
 </A>
<A NAME="marker=45092">
 </A>
Simultaneous Search</H3>
<P CLASS="Body">
<A NAME="pgfId=8490">
 </A>
If an anchor-first search is not selected, then a simultaneous search is performed by default.  All torsions are searched and/or minimized in concert.  The conformation search is performed prior to the orientation search, so each conformation is docked independently.  The simultaneous search technique is useful for conformation analysis and for constructing a chemical screen database (see <A HREF="Manual.12.html#40576" CLASS="XRef">
Chemical Screen on page 44</A>
).</P>
<DIV>
<H6 CLASS="Heading-5">
<A NAME="pgfId=46507">
 </A>
Constraining the search</H6>
<P CLASS="Body">
<A NAME="pgfId=46415">
 </A>
During a simultaneous search, dock performs an exhaustive or a random search, depending on the flexibility of the molecule and the value of <A HREF="Manual.19.html#16803" CLASS="XRef">
conformation_cutoff_factor</A>
.  The cutoff on the number of conformations generated for a molecule is calculated by <A HREF="Manual.f.html#37205" CLASS="XRef">
Equation 4</A>
.</P>
<DIV>
<H6 CLASS="Equation">
<A NAME="pgfId=46436">
 </A>
<A NAME="37205">
 </A>
<IMG SRC="Manual-11.gif" ALIGN="BASELINE">
Equation 4</H6>
<P CLASS="Body">
<A NAME="pgfId=46437">
 </A>
If the total conformations for a molecule is below Ncut an exhaustive search of all conformations is performed.  Otherwise, a random search of Ncut conformations is performed, in which torsion positions are selected randomly from the allowed positions.  This definition of Ncut was used so that the time demand grows linearly with the number of rotatable bonds, which is a fair compromise between the exponential growth of the total possible conformations and a fixed cutoff applied uniformly to all molecules.</P>
</DIV>
</DIV>
<DIV>
<H6 CLASS="Heading-5">
<A NAME="pgfId=46506">
 </A>
Caveat on backtracking procedure</H6>
<P CLASS="Body">
<A NAME="pgfId=46483">
 </A>
A backtracking procedure is used during the generation of each conformation.  If an internal  clash is detected for the torsion position of a rotatable bond within a partially-built structure, then another torsion value is attempted.  If no torsion values will work, then the procedure backtracks to the preceding rotatable bond and assigns a new torsion for it.  This procedure works fine when internal clashes can be resolved easily, but in the worst case, for a molecule with a huge number of rotatable bonds and a clash that cannot be resolved at the last rotatable bond, the procedure will be confounded and tend to consume cpu time.  For this reason, when processing a database of molecules, be sure to use a reasonable value of <A HREF="Manual.19.html#19540" CLASS="XRef">
flexible_bond_maximum</A>
 to discard overly flexible molecules.</P>
</DIV>
</DIV>
<DIV>
<H3 CLASS="Heading-4">
<A NAME="pgfId=5314">
 </A>
<A NAME="marker=11666">
 </A>
Torsion Minimization</H3>
<P CLASS="Body">
<A NAME="pgfId=6648">
 </A>
The torsion angles of rotatable bonds may be included in the <A NAME="marker=11667">
 </A>
<A HREF="Manual.10.html#23381" CLASS="XRef">
Score Optimization</A>
.  This provides for a much more efficient conformation search since fewer torsion positions need to be sampled.  Each torsion flagged for movement is assigned a simplex vertex along with the six rigid body degrees of freedom.  Only non-bonded interatomic terms are included in the scoring evaluation; no explicit torsion terms are included.  Therefore, only torsions flagged as minimizable in the <A NAME="marker=11804">
 </A>
<A HREF="Manual.4b.html#26494" CLASS="XRef">
flex.defn</A>
 file are included (e.g. double bonds are excluded by default).  When an <A NAME="marker=11803">
 </A>
<A HREF="Manual.f.html#37908" CLASS="XRef">
Anchor-First Search</A>
 is performed with segments from multiple layers being minimized, then inner torsions are assigned smaller initial torsion step values since perturbations in these torsions have a greater impact on conformation.</P>
<DIV>
<H6 CLASS="New-Page">
<A NAME="pgfId=11265">
 </A>
&nbsp;</H6>
</DIV>
</DIV>
<CENTER><P>