The X-LIGAND application automates the process of fitting ligand coordinates to electron density maps with little or no user intervention.
X-LIGAND was designed for the crystallographic protocol in which ligand binding experiments have successfully produced protein/ligand complexes that form isomorphous crystals with respect to the apo form of the protein. These ligand binding experiments result in isomorphous crystallographic data where the protein part of the model structure can be solved directly, or by molecular replacement, using the apo protein model. The ligand binding sites should be apparent within omit density maps from these experiments as connected density that corresponds in approximate size to that of the ligand. There will also be other electron density due to data and model errors.
This experimental protocol can be relatively rapid with many ligands being successfully bound and data collected in a short period of time. The result is that the model building process becomes the rate limiting step for the structure determination.
X-LIGAND is therefore designed to search for unsatisfied electron density (density containing no molecular coordinates), sort these in order of volume, and fit a ligand to these sites automatically. The application is also able to search conformation flexibility of a ligand by varying any rotatable bonds and fit these rotamers to density at a rate of more than a thousand per second. This entire process, including refinement, can be carried out quickly with X-LIGAND.
Because of the success rate of producing apparently correctly fitted molecular coordinates to electron density, X-LIGAND has been found useful for any general fitting of coordinates to omit/2Fo-Fc/fo-fc data, including small inorganic molecules (such as sulfate ions), polysaccharides, and polypeptides.
X-LIGAND requires a set of protein coordinates, one or more sets of ligand coordinates, a map, and symmetry of the crystal system. The map may either be of difference density or (n+1)fo-(n)fc as the program excludes all regions with model coordinates already present in density. If the ligand has rotatable bonds, the initial conformation of the ligand is not important.
Because X-LIGAND is designed to use multiple ligands that do not "see" each other (nonbond interactions), then all parts of the molecule for which you want to calculate nonbond interactions must be part of the first MSF. This is not normally a problem with molecules whose waters are not included in the file, because the solutions from the search are sorted by size, and hence the best solution is first.
The application uses a routine to determine the rotatable bonds within ligands, since this is necessary for all the refinement methods used in X-LIGAND. This routine is not based on atom typing and does not require the presence of hydrogens. This means that X-LIGAND is able to determine the correct number of rotatable degrees of freedom for any ligand regardless of the hydrogen mode of the molecule. Since the algorithm is based on bonding of the ligand and geometrical analysis of the structure, the ligand used in X-LIGAND must be close to its energy minimum. Since this is required for the application (as only torsional conformational space is explored), this does not represent a limitation in the application.
Rotatable bonds are rejected in a structure if:
1. The two central atoms are part of a ring.
2. The 2nd or 3rd atom is a terminating atom.
3. The two central atoms are able to form a partial double bond such as two sp2 carbon atoms.
4. The first or fourth atom is a hydrogen.
Since hydrogen atoms (in polar or all-atom models), have little electron density associated with their position, fitting hydrogen atoms increases the complexity of the search problem while adding little to the quality of the final structure. If you must place hydrogen atoms in the search, as with high resolution structures, then you can add torsions that define their position using the tool Define 1 torsion.
If the rotatable bonds are changed to include more, or remove some, then the application creates a file of the new definitions. It is not necessary to know the contents of this file, or edit the file, as there are tools within X-LIGAND and X-BUILD to edit this file using a graphical user interface. This file is of the format:
Title lineResidue_name atom1 atom2 atom3 atom4
For example, for a lysine and phenylalanine residue, the file could contain:
Rotatable bonds for the two amino acids lys and phe
LYS N CA C O
LYS N CA CB CG
LYS CA CB CG CD
LYS CB CG CD CE
LYS CG CD CE NZ
PHE N CA C O
PHE N CA CB CG
PHE CA CB CG CD1
The title line can be anything and is ignored, but must be present. The residue/atom definitions are free format and delimited by spaces.
Each line defines a rotatable bond for a residue. There can be as many as 100 rotatable bonds for a single residue, and as many residue definitions as required. The ligand must only have a single residue name. The residue name is a string of up to four letters and must be the same as this ligand's name in the MSF. The series of four atom names on each line defines the torsion angle; the rotatable bond lies between the second and third atom. Any lines in this file that lack either the residue name for the current ligand or atom names are ignored.
The protein molecule must be the first molecule in the Molecule Management table. If this is not so, close all the preceding MSF files and reopen them, using the append mode before you enter X-LIGAND. X-LIGAND aborts if the first molecule is not a protein.
A ligand to be fitted must be provided to X-LIGAND as a separate MSF file. Multiple ligands can be provided to X-LIGAND so that each different ligand is stored in different MSF files. Hence it is necessary that at least two MSF files must be open, displayed, and active to use the X-LIGAND, but more than two MSF files can be provided if different ligands need to be fitted to a protein. The protein coordinates must be in the first MSF file.
The X-LIGAND application generates a sphere of symmetry atoms around the working molecule so that symmetry ligand sites are not found as multiple sites. The symmetry atoms display as blue and the NCS atoms as red.
X-LIGAND first finds all sites around the molecule where the electron density is above a user-defined threshold value, but in which no model atoms occur. These sites are then sorted by size, with the largest region of density first. The ligand is placed at this largest site by the application, and if the ligand has no internal degrees of freedom, then it is necessary only to refine the results.
Once all the sites have been determined by the application, if there are several internal degrees of freedom in the ligand, possible conformations can be fitted to the density at the rate of about 2000/second. The conformations are searched by changing the rotatable bonds, where the rotatable bonds are determined automatically if the molecule has been built in the 2D builder, or provided explicitly as an external file. X-LIGAND automatically determines the precision at which rotatable bonds need to be searched to efficiently give good results. The largest ligand so far tested had 16 rotatable bonds; it took about 10 minutes to solve (about 1,000,000 conformations searched).
X-LIGAND weights the search precision of each torsion depending on the effect each torsion has on the overall shape of the molecule. Torsions in the middle of the molecule will be searched to the maximum precision of 1°, while those torsions that only affect the placement of 1 or 2 atoms may only be searched at a precision of 180°. This weighting scheme allows the application to be computationally efficient, and prevents many solutions with almost identical coordinates.
Once the application has searched the conformations and found a good solution to the fitting problem, it is possible to carry out manual changes to improve the fit. These can be carried out at any time, and may be necessary if the ligand is particularly flexible (more than 10 internal degrees of freedom).
The ligand can now be refined using torsion angle real space refinement to the electron density. This has a very high radius of convergence (about 2 Å), and will refine the ligand regardless of errors remaining from the search. The refined ligand can be saved to a new MSF at any time.
For a rigid molecule, the program will immediately fit the ligand to a site. You need only refine the position.
The conformation search tries to simultaneously fit the position, orientation, and rotatable bonds of the ligand, and keeps the results that fit the density the best. If you have a particularly flexible ligand (more than ten torsions), then the program will probably require an omit map so that you do not get fits to other density. For more rigid molecules, you can use a 2fo-fc map, since X-LIGAND checks each site for nonbonds. The refine option is very powerful, and will often fit something a long way from it minima.
Initially, only the first three tools are unmasked, (plus Exit), as nothing can be calculated until the sites have been searched for. Once the ligand sites are found, other tools become active. If there are no internal degrees of freedom, the conformation search and the sub-tools under it will not be unmasked, since they are not relevant for this type of ligand. If you wish to change the torsion list, the tool to add/del 1 torsion can be used to edit the list of torsions at any time, and the Select new ligand tool used to re-read the contents of this file and reset all the necessary options. If you have only one ligand, this process occurs without a prompt, otherwise a dialog box prompts you to select the ligand again.
This searches for regions of density that have no atoms and returns the sites, sorted by size. The biggest region of density is the first site. The ligand site size is defined by volume of density that is connected, above the defined threshold (default=1.5 sigma), and not overlapped with the atoms in the first MSF. This will take about 5-10 sec. The program places the ligand, in its current conformation, in approximately the best orientation at the first site.
This tool adjusts the volume of the electron density to be searched to be the same as the volume of the ligand. This works by adjusting the sigma value used in flood fill until the density volume found matches the volume of the ligand. The initial value for the sigma level is defined within the X-LIGAND | Change search setting...| Search threshold tool. In general, it appears that setting this to quite a high value (3.0) is best. The Adjust site volume tool then searches down to find the correct extent of electron density for the ligand to fit into. Occasionally, there is a problem with the tool not finding the correct volume if the initial value is left too low.
This opens a dialog of options that allows basic search parameters to be modified.
This sets the map level (in sigma) at which the search for possible ligand sites is made in the map. A lower threshold will result in more sites being found, and will also increase the extent of the volume of each site that does not overlap the protein. A lower threshold will often result in the overlap of sites, giving fewer, larger sites. This is the only parameter that needs to be set by you. The default is appropriate for searching 2fo-fc maps.
This is automatically set by the program. If the number of internal degrees of freedom results in the total number of searched conformations exceeding 50,000, then a Random search is set, otherwise a Grid search is set. You can override these defaults by selecting the mode you want.
Distance and positional restraints are supported to increase the success rate of the conformational search during ligand fitting.
Symmetry atoms can be picked and the information displayed.
This option allows you to eliminate electron density in the ligand binding site that overlaps with the protein atoms or their symmetry equivalents.
Defines the type of non-bonding interactions between the receptor molecule and ligand. The following interactions are supported:
This allows you to turn on or off the option of having flexible partial bonds in the ligand conformational search. For example, the peptide bond.
To place the ligand in the density initially, X-LIGAND compares gyration radii and moments of inertia for the density and the ligand to determine the initial position and orientation of the ligand in the density. There is a new option to use third moments of inertia for the shape-matching. If No 3rd moment is chosen, the second moments of inertial are chosen, as in the earlier release of QUANTA.
Option to apply Z-weighting restraints to the ligand atoms based on their atomic number. This option is helpful if the ligand contains heavier atoms such as Hg or S. The ligand-placement algorithm will weight both the heavier atoms in the ligand, as well as the regions in the ligand electron density volume.
This option allows you to perform or skip regularization on the ligand geometry before and after conformational searches.
In a grid search, this parameter forces an upper bound to the number of grid values to search. The program automatically determines a grid search resolution for each torsion, so if the product of all torsions to search exceeds this number, the grid step sizes will be increased until the total number of search conformations is less than this value.
This parameter defines the length of the Monte Carlo search.
This option defines whether nonbond interactions between ligand atoms are to be included (none or self) and allows you to turn on or off the calculation of nonbond interactions with the protein (full). For most ligands and electron densities, this energy term is not needed for a reasonable fit, particularly when the electron density volume is close to that of the ligand volume.
When the electron density volume is smaller than the ligand, or the ligand is to be fitted into a small void of the protein, then nonbond terms are needed between the ligand atoms to aid the fitting process. The ligand will fold up onto itself if these terms are not calculated.
This energy term can be turned on and off, because there is a time penalty for this energy term, the calculation usually being about 3 times slower. Nonbond terms between the ligand and any other visible molecule are always calculated, since they are pre-calculated before the search and thus provide no time penalty within the calculation.
The default maximum number of torsions to search for a grid search is 12. If there are more than 12 torsions to search, X-LIGAND will not search the least important torsions.
This parameter sets the radius of displayed map about the ligand.
X-LIGAND sets this parameter to cover the entire molecule of interest plus 5 Å beyond the most distant atom. Use this option to change the scope of the search algorithm.
The tool to position search center allows a particular region of the map to be focused on by setting the center of the map search to a particular point in space, and then using a smaller calculation radius (X-LIGAND | Changesearch setting... | Radii / Calculation radius). This allows faster use of the application on very large molecules, but is probably not necessary for most problems. On selecting this tool, a pointer will appear at the screen center (the current search center), and can be moved with the dials. A new menu will open, allowing you to accept or reject the new position. See Using the 3D Pointer, for detailed information on the Pointer palette.
If Accept position is selected from the Pointer palette, the search center will move to this position, and symmetry recalculated based on this center and the calculation radius. If Quit is selected, then no change is made.
This tool provides an easy method to turn the map display on and off from the X-LIGAND menu. When the maps are not displayed, the ligand site can be observed more easily, and on machines with slower graphics, refinement and ligand editing will proceed more quickly because the display will be redrawn more quickly.
This tool resets the position of the ligand to that of the original ligand on entry into the X-LIGAND module. This is needed when scoring the fit of a particular ligand conformation to the density.
This moves the display to the next potential ligand site and places the ligand, in its current conformation, in the site in its best orientation.
This moves the display to the previous potential ligand site and places the ligand, in its current conformation, in the site in its best orientation.
This allows you to go to a particular site directly rather than navigating with the Next site/Previous site options.
If the ligand contains internal degrees of freedom such as rotatable bonds, then it is possible to search conformations of the ligand as a function of these torsion angles. The program can search up to 100 rotatable bonds simultaneously.
During the search, the program automatically weights each rotatable bond according to the effect rotating the bond would have on the structure. For bonds that will significantly affect the structure, the search step will be 1°. For a rotatable bond that has very little effect on the structure, the search angle will be 180°.
If the total number of search conformations exceeds 50,000, the application selects the monte carlo search method, which is stopped only by an aborting mouse click or when the search time limit expires after 10 minutes. If the number of conformations to search is less than 50,000, the application carries out an exhaustive search, normally in less than one minute. In this case it would be normal to allow the application to complete the search, but you can abort it with a mouse click if you need to. During the search process, the current best solution is displayed, so if you see a good solution, you can stop the search and this best solution becomes the first of the best 20 displayed.
This tool refines all the 20 conformations found by the Search conformations tool and then orders the conformations on the basis of the scoring function.
The program, on completing the search, lists the 20 best solutions in order of density fit. You will often find alternative conformers in this list. You can look through the list using the tools next conformation and previous conformation. Next conformation will show the next best fit, unless the current fit is the last one, in which case the next fit is the first one.
The program, on completing the search, lists the 20 best solutions in order of density fit. You will often find alternative conformers in this list. You can look through the list using the tools next conformation and previous conformation. Previous conformation will show the previous best fit, unless the current fit is the first one, in which case the previous fit is the last one.
This tool will show the best 20 conformations found by the conformation search. This can show how the conformations cluster and may indicate a single well-fitted conformation when all 20 are very similar, or a weakly restrained conformation when all 20 conformations are similar but show significant variation, or that there are multiple conformations.
Note that when this mode is highlighted then it is not possible to refine or save the coordinates as the applications does not know to which structure the operation refers. Selecting this tool again will turn of the display of all the best solutions and return to the previous displayed solution.
If the search for conformation results in less than 20 solutions, then only those found will be displayed.
This tool allows you to turn off the positional fitting of a ligand. This is useful when a monomer of a polymer is being fitted to density for each monomeric site. To use this option, move the ligand manually to the center of the required site, then fix the origin. When a conformation search is subsequently carried out, only the orientation and internal degrees of freedom are searched for that ligand, thereby fitting the ligand only to the position required. This tool does not make the search proceed any faster.
This tool allows you to fix one or more atoms in the ligand molecule during the conformational search.
This tool resets the atoms fixed by the previous tool.
This tool fits the portions of the ligand with better density (based on shape-matching) before fitting the rest of the ligand molecule. This option reduces computation time.
After a ligand is fit with this command, one or more previously unconstrained atoms may end up being fixed and several torsions may also be inactivated. Repeated ...fit in stages sessions can proceed from that state, but you may sometimes prefer to reactivate some or all torsions. In that case, the constraints from atoms and torsions should be manually removed before repeating ...fit in stages. The commands ...Clear fixed atoms and Set active torsions should then both be used.
This tool, which is a copy of the one on the X-BUILD | build atoms palette, flips a selected torsion within a ligand by 180 degrees. The selected torsion within the ligand must be rotatable or nothing will be done.
This gives a menu with all the active torsions highlighted. The torsions present in the molecule are shown on the ligand molecule for reference. Initially, all the torsions are set as active (denoted by a check mark before the torsion), and can be inactivated by selecting the tool for a torsion. An inactive torsion is not searched or refined.
When a flexible ligand molecule is fitted to electron density using the X-LIGAND | Search conformations tool, it will be found that the central core of the ligand will be fitted to the density first. As the search continues, torsions will be refined so that larger parts of the structure will be fitted around the central core of the ligand molecule with peripheral parts of the ligand poorly fitted. It is therefore possible with this tool to inactivate the torsions that no longer need to be refined by the search procedure because this part of the ligand is already fitted. This will improve the radius of convergence for fitting regions of the ligand not yet fitted.
Allows you to specify and remove torsions manually by picking a single bond. When selected, the tool will label all the current rotatable bonds (either calculated, or from the file lig.rot), with the torsion number. It is possible to pick a bond that is either already defined as a torsion, which will be subsequently deleted, or a new bond that will be added to the list of torsions, if valid. An invalid torsion is one that is part of a ring or is a terminating bond. It is possible to add a torsion between atoms not normally recognized as rotatable, such as a peptide bond or where the 1st/4th atom is a hydrogen atom. After picking a bond, the tool exits. If you want to continue adding torsions, you should use the same tool again.
Once this tool has been used to modify the rotatable bonds in a ligand, the application creates or modifies the lig.rot file, lig.rot, which contains the definitions for rotatable bonds. This file is then used in any subsequent analysis of the ligand. Note, other definitions for different residues are retained in the lig.rot file.
This allows you to manually edit the torsion angles of the ligand, up to a maximum of 100 torsions. If there are more than seven rotatable bonds, then the eighth dial allows you to toggle up and down the list of rotatable bonds. If you have no torsions, this option does nothing.
This allows you manually move the ligand position and orientation. The dials will change to allow movement of the ligand in xyz and rotated about the xyz axes.
This does real space refinement of the current ligand conformation's torsion angles in the current site. When you select the refine tool, the position, orientation, and all defined torsion angles are refined to the electron density by least squares.
This ligand scoring tool takes the current ligand conformation and assesses how well the ligand fits the density. The larger the score, the better the fit.
This creates a new MSF of the current coordinates. The application will open a file browser and prompts for a filename for the new ligand position and conformation. The option SAVE saves the current ligand coordinates, while CANCEL aborts the save and does nothing.
This tool allows a different ligand to be selected from the currently open MSF files. If more than one ligand was open on entry to X-LIGAND then a dialog box will appear, allowing a new ligand to be selected. If Select is chosen, this will become the new active ligand and will be fitted, and sourced in subsequent parts of X-LIGAND. If Cancel is picked, no action is taken. This option resets all torsions and recalculates the step size for the search, and checks the search method to use.
This tool is used to carry out multiple ligand fits to a single site. For the current site it is possible to:
A dialog box is opened that has a file browser and 2 sets of options. A list of skeleton data base files are shown (*.ind) as default. A skeleton file can be chosen to carry out a database search.
The first option list is for the file type. The default is the list of skeleton files (*.ind). If the second option is chosen then all the MSF files in the directory are listed with the file name *001.msf. This option allows a small number of ligands to be placed into multiple MSF files with names *nnn.msf where nnn is a number from 001 to 999. This is designed for simple access to a small number of ligands. Each MSF file must only have 1 ligand in.
The second option defines what data is to be returned. The default is to return the best fit ligand, all ligands currently open (molecules 2 and higher) will be closed, and on completion the best fit ligand is opened. The "Best 20" option will return the best 20 ligands, and on completion of the search the best 20 (or the number of ligands in the databank) will be opened and displayed with the best fit ligand as molecule 2 and the worst of the 20 ligands as molecule 21.
If cancel is picked from the dialog box nothing is done. If OK is picked from the dialog box then the search is carried out.
The Exit tool will exit from the X-LIGAND application and restore all the previous atom selections. If any new ligand positions have been saved while in X-LIGAND (X-LIGAND | Save ligand to MSF), then these will be opened and be referenced in the molecular management table. The position of the search ligand will be that on entry to the X-LIGAND unless a new ligand was selected while in X-LIGAND (X-LIGAND | Select new ligand) when it will appear at the site on changing the active ligand.
The Exit tool in X-LIGAND will not save the current ligand position; these new fitted sites are saved using the tool X-LIGAND | Save ligand to MSF.