Align-it™

Citing Align-it

Publication of data and results, partly or entirely generated using Align-it, should cite the original reference of Align-it™:

Taminau, J.; Thijs, G.; De Winter, H. (2008) ‘Pharao: pharmacophore alignment and optimization’, J. Mol. Graph. Model. 27, 161-169.

Version

Align-it 1.0.4

Feedback

Join our Google groups community to talk about inconsistencies, errors, raise questions or to make suggestions for improvement.

Introduction

Align-it™ is a tool to align molecules according their pharmacophores. A pharmacophore is an abstract concept based on the specific interactions that have been observed in drug-receptor interactions: hydrogen bonding, charge transfer, electrostatic and hydrophobic interactions. Molecular modeling and/or screening based on pharmacophore similarities have proven to be an important and useful method in drug discovery.

The functionality of Align-it™ consists mainly of two parts. The first functionality consists of the generation of pharmacophores from molecules (see the section on generation of pharmacophores below). Second, pairs of pharmacophores can be aligned (see the section on aligning pharmacophores) and the resulting score is calculated from the volume overlap resulting from the alignments.

For the representation of a pharmacophore point, a 3-dimensional Gaussian volume is used in which the volume is defined by its center and spread or sigma.

Since alignment methods are dependent on the molecular orientation and position, they tend to reflect a combinatorial problem that sometimes results in extensive computation times. Several approaches are introduced within Align-it™ to handle this problem and this makes Align-it™ a screening tool that is sufficiently fast. The alignment as implement in Align-it™ is called a ‘rigid alignment’, meaning that no flexibility of the input structures is assumed and that the program always works with one - fixed - conformation. To obtain additional conformations of a molecule, external software should be used in a preprocessing step.

In the following section, implementation details are given in order to provide some insight in the working of Align-it™. In the usage section, a detailed explanation of the command line parameters and functions is given.

Implementation details

Pharmacophores

Concept

A pharmacophore is described as an ensemble of functional groups, or structural features, with a defined geometry. In Align-it™ a pharmacophore is represented as a set of pharmacophore points, whereby each pharmacophore point is characterized with the following properties:

  • the type of the functional group;
  • the center of the point;
  • the spread (\(\alpha\));
  • the normal, if applicable.

Each pharmacophore point is modeled as a 3-dimensional spherical Gaussian volume represented by its center (coordinate) and spread (\(\alpha\)). The definition of a Gaussian volume is given as follows:

\[V_a = \int p \exp(-\alpha (m - r)^2) dr = p \sqrt {(\frac{\pi}{\alpha})^3}\]

with \(V_a\) being the atomic Gaussian volume, p the normalization constant to scale the total volume to a level that is in relation to atomic volumes, m being the center of the Gaussian, and r being the distance variable that is integrated.

The coordinate m of a pharmacophore point defines the position in space. All pharmacophore points have a position in space. Each pharmacophore point is also characterized by \(\alpha\) that defines the spread of the Gaussian volume in space. \(\alpha\) is chosen inverse proportional to the square root of the radius.

Each pharmacophore point is characterized by a functional type. These functional types are considered to be important in the selective binding of molecules. Each functional group is labeled with a four-lettered code and the possibilities as implemented within Align-it™ are given in the table below.

Some of the pharmacophore points also have a direction as defined by its normal. The normal is a vector originating from the center of the pharmacophore point. It is optional to include this information during alignment and scoring. The rationale for the use of a normal in the alignment is that, for instance, a hydrogen bond acceptor works to the outside of the molecule, and an aromatic ring is a planar structure that has an orientation in space. This spatial orientation is not modeled as such by the Gaussian volume, hence the use of the normal is to take this orientation into account.

Code Description α Normal Hybrid
AROM Aromatic ring 0.7 Yes No
HDON Hydrogen bond donor 1.0 Yes No
HACC Hydrogen bond acceptor 1.0 Yes No
LIPO Lipophilic region 0.7 No No
POSC Positive charge center 1.0 No No
NEGC Negative charge center 1.0 No No
HYBH Hydrogen bond donor and acceptor 1.0 Yes Yes
HYBL Aromatic and lipophilic 0.7 No Yes
EXCL Exclusion sphere 1.7 No No

Format

Once generated, pharmacophores can be written to a file using a special whitespace-delimited format. This way, pharmacophores of molecules can be stored and used for screening or mapping without generating this information each time again. It is recommended to use the .phar extension for pharmacophore files.

The following format is used in Align-it™ for reading and writing pharmacophores:

name
CODE Cx Cy Cz α norm Nx Ny Nz
...
CODE Cx Cy Cz α norm Nx Ny Nz
$$$$

Every pharmacophore starts with a variable name, which is used to identify the pharmacophore. In principle, the name of the pharmacophore is set identical to the title of the molecule of which the pharmacophore is calculated. Then for each pharmacophore point a new line is used, containing the following information:

  • CODE is one of the nine codes listed in the table above;
  • Cx, Cy and Cz are the coordinates of the pharmacophore point;
  • \(\alpha\) is the spread of the Gaussian;
  • norm is a Boolean parameter (1 or 0) indicating whether this particular point contains normal information;
  • Nx, Ny and Nz are the coordinates of the normal. For pharmacophore points with no normal information, these three data points are set to 0.

The end of the pharmacophore is indicated with four dollar signs. This way, a file can contain multiple pharmacophores. Lines starting with a # symbol are considered comment lines and are skipped during parsing of a pharmacophore file.

This human-readable format enables the manual modification of a pharmacophore set. To remove a pharmacophore point from a pharmacophore, it is sufficient to remove the corresponding line in the file.

Generating pharmacophore points

Aromatic rings

The generation of aromatic ring pharmacophore points, or AROM points, includes ring detection and aromaticity detection.

Ring systems containing multiple aromatic rings will be converted into multiple AROM points. Figure 1 illustrates this for the molecule naphthalene, consisting of a ring system with two benzene rings:

Figure 1

Figure 1. Visualization of the two generated AROM points for naphthalene. Both points are shown as yellow spheres. The normals are not shown.

The position of the AROM point is the center of the ring it represents. AROM points also contain a normal as extra information. This normal indicates the orientation of the aromatic ring and is placed perpendicular on the plane formed by the ring. Because its sole purpose is to indicate the orientation of the plane, the normal is always a unit vector with length 1 Å.

If the angle between two normal vectors is zero, then two corresponding ring planes are parallel to each other. The value of this angle can act as a penalty when comparing two AROM points to each other.

Hydrogen bond donors

The generation of hydrogen bond donor pharmacophore points, or HDON points, is based on topological information according a simple procedure. For an atom to be labeled as a hydrogen bond donor, the atom should fulfill the following conditions:

  • Only nitrogen or oxygen atoms;
  • Formal charge is not negative;
  • At least one attached hydrogen atom.

Note

There is no need to have explicit hydrogen atoms being added to the molecule before extracting the pharmacophore points, as the program is using implicit hydrogen bond counts.

The center of the HDON point is the position of the heavy atom that is labeled as a valid hydrogen bond donor. Hydrogen bond donor pharmacophore points are also characterized by normal information. The direction of this normal is calculated from the average position of all the non-hydrogen atoms that are bound to the hydrogen bond donor atom, shifted to a length of 1 Å and projected along this vector to the other side of the hydrogen bond donor atom (Figure 2). The position of the hydrogen atom is not taken into account for the calculation of the normal.

Figure 2

Figure 2. Illustration of the procedure to position the normal on a hydrogen bond donor pharmacophore point as shown for a hydrogen bond donor atom connected to a single heavy atom (a) or to two heavy atoms (b). The hydrogen bond donor atom is colored red, the associated normal point light blue, and the attached atoms gray. A similar procedure is used to calculate the normals of the hydrogen bond acceptor pharmacophore points.

Hydrogen bond acceptors

The generation of hydrogen bond acceptor points, or HACC points, is less straightforward than the generation of HDON points. A hydrogen bond acceptor needs to fulfill four conditions:

  • Only nitrogen or oxygen atoms;
  • Formal charge not positive;
  • At least one localized lone pair;
  • Atom is accessible.

These conditions, which will be described in more detail below, are based on the work of Greene and coworkers [1].

In order to determine condition three - the presence of at least one localized lone pair - only nitrogen atoms have to be validated for the presence of localized lone pair electrons. Some simple heuristic rules have been implemented to validate this condition. A nitrogen has no localized lone pair electrons if the nitrogen obeys one of the following patterns:

  • N is part of an aromatic ring and has three bonds attached to it (e.g. pyrrole);
  • N-S=O (e.g. sulfonamide);
  • N-C=X with X equal to N, O or S (e.g. peptide bond);
  • N is adjacent to aromatic ring and has three bonds attached to it (e.g. aniline).

All other nitrogen atoms are flagged to have at least one localized lone pair.

The fourth condition in the definition of a hydrogen bond acceptor - the accessibility of the atom - is somewhat more difficult to calculate. Accessibility means that there is enough space for a putative hydrogen atom to form a hydrogen bond without forming a steric clash with any of the other atoms of the molecule.

This accessibility is calculated by placing a sphere around the putative hydrogen bond acceptor atom with a radius of 1.8 Å, thereby mimicking the possible locations where a hydrogen atom can be localized in theory. Subsequently a number of points are sampled on this sphere and for every point on this sphere it is verified whether a collision with any of the neighboring atoms might occur. If at least 2% of the points are labeled as ‘non-colliding’, the putative hydrogen bond acceptor atom is labeled as being accessible.

By imposing the third and fourth condition as additional criteria for the determination of a hydrogen bond acceptor pharmacophore point, the number of hydrogen bond acceptors are significantly reduced (Figure 3).

Figure 3

Figure 3. Illustration of hydrogen bond acceptor pharmacophore points. Only one HACC point was generated and is shown as a yellow sphere. The molecule contains three nitrogen atoms that could serve as hydrogen bond acceptor pharmacophore centers, but only the right-most nitrogen satisfies all four constraints and therefore gets labeled as a hydrogen bond acceptor. The normal of the point is not shown.

The normal of the hydrogen acceptor pharmacophore point is calculated in an identical manner as for the calculation of the normals of the hydrogen bond donor pharmacophore points (Figure 2).

Lipophilic spots

To generate lipophilic pharmacophore points, or LIPO points, a procedure as described below is used.

First, each atom is assigned a lipophilic contribution value. This value is the product of a topology-dependent term t and an accessible surface fraction s. The term t is obtained from a number of heuristic rules that are listed in the table below. The fraction s, representing the accessibility of an atom, is calculated using a method similar to the method as described for the calculation of hydrogen bond acceptors. For example, a carbon atom with an accessibility of 80% and located three bonds away from double bonded oxygen will have a lipophilic contribution of 0.48 (s = 0.8, t = 0.6):

Category     f    Description
--------   ----   ------------------------------------------------------------------
   1       0.00   N, O or H
   2       0.00   S in SH
   3       0.00   ≤ 2 bonds away from charged atom
   4       0.00   ≤ 2 bonds away from OH or NH with no delocalized electrons
   5       0.00   ≤ 1 bond away from SH with no delocalized electrons
   6       0.00   ≤ 2 bonds away from O with double bond
   7       0.00   ≤ 1 bond away from S with valence > 2
   8       0.00   S with double bond
   9       0.60   3 bonds away from O with double bond
  10       0.60   2 bonds away from S with valence > 2
  11       0.60   1 bond away from S with double bond
  12       0.00   ≥ 2 instances of any of the previous three conditions (cat 9-11)
  13       0.25   1 neighboring O or N with no delocalized electrons
  14       0.00   > 1 neighboring O or N with no delocalized electrons
  15       1.00   Not belonging to any of the previous conditions (cat 1-14)

After having assigned the lipophilic contribution to each atom, the second step is to group atoms into regions or spots. The procedure to group atoms into spots is illustrated in Figure 4 below, and is based on a number of rules:

  • Atoms togther in a ring of size 7 or less form a group (Figure 4a).
  • Atoms connected to three or more atoms, and those neighbors that are not bonded to any other non-hydrogen atom, form a group (Figure 4b).
  • The remaining of the atoms (the chains) also form groups (Figure 4c).
Figure 4

Figure 4. Schematic representation of procedure to group atoms into spots. This example molecule contains four spots.

During the third and final step, for each of the identified spots the total lipophilic contribution is calculated as the summation of the contributions of every atom belonging to that spot. If this value exceeds a predefined threshold, a LIPO pharmacophore point is created with the center being set to the center of the spot. The threshold value is set to 9.87, which is half of the lipophilic contribution of an exposed methyl carbon terminating a carbon chain [1].

Charge centers

The formal charges on the atoms of the molecule are used for the generation of charge center pharmacophore points. Atoms with a positive formal charge will correspond with a positive charge center pharmacophore point, or POSC point, and atoms with a negative formal charge will define the position of a negative charge center pharmacophore point or NEGC point.

The position of the POSC and NEGC points coincides with the position of the atom carrying the formal charge.

Hybrid lipophilic centers

Hybrid lipophilic pharmacophores HYBL are generated by merging proximate LIPO and AROM points together. In order for these to be merged, the distance between the two respective centers should be less than 1.0 Å. The center coordinates of the new point are calculated by taking the average of the two original centers. When hybrid lipophilic centers are requested, all LIPO and all AROM points are renamed to HYBL. After merging and renaming, the normal information of the original aromatic centers is disguarded.

To summarize, generation of HYBL points is done as follows:

  • Isolated AROM points are renamed to HYBL and their normal information is disguarded;
  • Isolated LIPO points are renamed to HYBL;
  • Proximate AROM and LIPO points are merged into a HYBL single point and the normal information of the original AROM point is removed. The new coordinates are calculated as the average of the original coordinates.

Hybrid hydrogen donors and acceptor centers

Hybrid hydrogen acceptor/donor pharmacophores HYBH are generated by merging together HACC and HDON points that are located on the same atom. In order for these to be merged, the distance between the two respective centers should be less than 0.00001 Å. The resulting new type is set to HYBH.

After merging, the normal of the new center is calculated by taking the average location of the two original normals.

Merging pharmacophore points

Because of the combinatorial nature of the feature mapping (see below), extended sets of pharmacophore points can lead to extensive computational times. A possible solution to circumvent this problem is to merge neighboring pharmacophore points of the same category, as is illustrated in Figure 5.

Figure 5

Figure 5. Schematic representation of the merging process. A pharmacophore consisting of six points is reduced to a new pharmacophore consisting of only three points.

Pharmacophore points are considered to be ‘neighbours’ if their overlap volume exceeds a threshold value of 0.075. The spread \(\alpha\) of the resulting pharmacophore point is set to 70% of the sum of all the original \(\alpha\) values. A merged pharmacophore point does not contain normal information.

Aligning pharmacophores

Problem situation

Quantification of the similarity between two pharmacophores can be computed from the overlap volume of the Gaussian volumes of the respective pharmacophores. The principle is to identify the subset of matching functional groups in each pharmacophore that gives the largest overlap. The procedure finds its roots in the work of Grant and Pickup [2], where the volume overlap between two molecules is computed from a Gaussian description of the atomic volumes. In Align-it™ this approach is translated into the overlap of pharmacophore points.

The procedure to compute the volume overlap between two pharmacophores is implemented in a two-step approach. During the first step, a list of all feasible combinations of overlapping pharmacophore points is generated. In the second step, the corresponding features are then aligned with each other using an optimization algorithm. The combination of features that gives the maximal volume overlap is retained to give the matching score.

Feature mapping

To compute the overlap between a pair of pharmacophores, the first step is to define the points from the first pharmacophore (A) that can be mapped onto the points from the second pharmacophore (B). A mapping of two pharmacophores consists of a list of points from A and B in which corresponding points have a compatible functional group and the internal distance between the correspodning points lies within a given range requirement. This range, as defined by the parameter \(\epsilon\), controls the overlap feasibility of a given combination of pharmacophore points.

The procedure starts by generating a list of all feasible feature pairs. First, two points from pharmacophore A are selected and the distance between these points is calculated. Next, two points with matching features and distance are selected from B. Subsequently, the first points of both couples are overlaid and the relative volume overlap between the second pair of points is computed according:

$$ \frac{V_o}{V_a + V_b + V_o} $$

in which \(V_a\) and \(V_b\) represent the volume of pharmacophore point a and b, respectively, and \(V_o\) the calculated absolute overlap volume. If this relative volume overlap is larger or equal than 1.0 - \(\epsilon\), the combination of the two pairs is set to be feasible. This is illustrated in Figure 6.

Figure 6

Figure 6. Illustration of the \(\epsilon\) parameter. Two subsets of corresponding pharmacophore points are selected (black and blue). The first points are placed on top of each other (left sphere). The relative volume overlap between the other spheres should be larger than \(1.0 - \epsilon\). From this it implies that a smaller \(\epsilon\) implements a more stringent feasibility criterion.

When \(\epsilon\) is set equal to 1.0, no limit on the minimal required amount of overlap between both pairs of pharmacophore points is imposed. Smaller values of \(\epsilon\) lead to a more stringent overlap criterion.

Once the initial list of feasible pairs is constructed, they can be combined into larger combinations. This process is combinatorial in nature and the number of possible combinations grows rapidly with the number of pharmacophore points in both pharmacophores. The choice for a stringent \(\epsilon\) value should aid in limiting the number of feasible combinations.

Alignment phase

Given the set of all feasible combinations, the one that gives the largest volume overlap is searched for. For every potential combination, the procedure starts by translating the A pharmacophore subset such that its geometric center overlaps with the geometric center of the B pharmacophore subset. Next, using a combination of gradient-ascent and rigid-body rotation, the maximal volume overlap is determined. Details of the methodology are described in reference [3] and the manual of our shape-it tool.

The alignment procedure starts with the combinations that are largest in terms of the number of matching pharmacophore points. Subsequently, smaller combinations are processed until the maximum score so far is larger than the theoretical maximum score any smaller combination could achieve, based on the underlying rationale that the maximum achievable volume overlap is limited by the number of features to align.

Alignment scores

Similarity between the pharmacophores A and B can be calculated using three different measures:

$$ \text{TANIMOTO} = \frac{V_O}{V_A + V_B - V_O} $$ $$ \text{TVERSKY_A} = \frac{V_O}{V_A} $$ $$ \text{TVERSKY_B} = \frac{V_O}{V_B} $$

with \(V_O\) being the maximum volume overlap between both pharmacophores; \(V_A\) the volume of pharmacophore A; and \(V_B\) the volume of pharmacophore B. The TANIMOTO measure is well known from bit vector comparison and is the default measure in Align-it™ to score similarity between pharmacophores.

Since the focus of the Align-it™ tool lies mainly in database searching experiments with a single reference pharmacophore as query, the similarity measures can be rewritten to reflect this:

$$ \text{TANIMOTO} = \frac{V_O}{V_{ref} + V_{db} - V_O} $$ $$ \text{TVERSKY_REF} = \frac{V_O}{V_{ref}} $$ $$ \text{TVERSKY_DB} = \frac{V_O}{V_{db}} $$

with \(V_{ref}\) the volume of the reference pharmacophore, and \(V_{db}\) the volume of the database pharmacophore. The TVERSKY_REF measure is primarily intended to identify database compounds with a pharmacophore that is a superset of the reference pharmacophore, while the TVERSKY_DB measure has its use in identifying database compounds having a pharmacophore that is subset of the reference pharmacophore.

All three metrics return a score between 0 and 1.

Command line usage

General

-h, --help
[OPTIONAL] Help on the use of Align-it™ is provided.
-q, --quiet
[OPTIONAL] If this parameter is provided, no output, progress or warnings are written out by the program.
--info <option>
[OPTIONAL] With this option the user can get detailed information for each <option> listed below.

For example, to get some information about the --dbase option, use:

> align-it --info dbase

or:

> align-it --info d

The <option> argument to --info is required, otherwise an error is written out:

> align-it --info
**MainError**  unknown command line option

Input

By default the format of input molecule files is determined from the extension of those files. Align-it™ supports all file types that are supported by OpenBabel. A pharmacophore file is specified with the .phar extension.

-r, --reference <file>
[OPTIONAL] This command line option defines the reference structure that will be used to screen and/or align the database molecules against. This option is not required, and when not given then the database will only be converted into pharmacophores without screening. By default the format is deduced from the extension of the file but this format can be defined explicitly with the --refType option. The <file> argument is required; if not provided then an error is written out.
--refType <type>
[OPTIONAL] With this option the format of the reference input file can be specified explicitly. The <type> argument is required but is case-insensitive. If this --refType option is not provided then the format of the reference file is deduced from its file extension. Allowed <type> argument keywords are those as understood by OpenBabel. Complementary to those types as defined by OpenBabel, an additional phar keyword, specific to Align-it™, defines that the reference input file will be in a specific pharmacophore format with precomputed pharmacophore points. If this specific pharmacophore format is not used, the program will automatically generate a pharmacophore from the reference using the procedure as described in the section on generating pharmacophore points. The cpu-time that is needed for this generation step is negligible compared to the cpu-time that is required for alignment.

Tip

To get a list of all file types that are understood by OpenBabel, type:

> obabel -L formats
-d, --dbase <file>
[REQUIRED] Defines the database of molecules that will be used to screen. This option is required. By default the format is deduced from the file extension but it can also be defined explicitly with the --dbType option.
--dbType <type>
[OPTIONAL] With this option the format of the database input file can be specified explicitly. The <type> argument is required but is case-insensitive. If this --dbType option is not provided then the format of the database file is deduced from its file extension. Allowed <type> argument keywords are those as understood by OpenBabel. Complementary to those types as defined by OpenBabel, an additional phar keyword, specific to Align-it™, defines that the database input file will be in a specific pharmacophore format with precomputed pharmacophore points. If this specific pharmacophore format is not used, the program will automatically generate a pharmacophore from each molecule in the database using the procedure as described in the section on generating pharmacophore points. The cpu-time that is needed for this generation step is negligible compared to the cpu-time that is required for alignment.

Output

-p, --pharmacophore <file>
[REQUIRED] The aligned pharmacophores of the structures in the input database are written to this file. The spatial position of these pharmacophores will not correspond to the original structures because they are aligned with respect to the reference input molecule and therefore can have a different orientation. Moreover, only the points that are used in the alignment are written out. If there is not a reference structure defined, or no alignment has taken place, then the complete pharmacophore is written out. This file is written in the specific farmacophore format.
-o, --out <file>
[OPTIONAL] The aligned database structures are written to this file. By default the format is deduced from the file extension but it can also be defined explicitly with the --outType option.
--outType <type>
[OPTIONAL] With this option the format of the molecular output file can be specified explicitly. The <type> argument is required but is case-insensitive. If this --outType option is not provided then the format of the molecular output file is deduced from its file extension. Allowed <type> argument keywords are those as understood by OpenBabel.
--cutOff <double>
[OPTIONAL] This value should be between 0 and 1 and only structures with a score larger than this cutoff will be written to the files defined by the --out, --scores and --pharmacophore options. The --rankby option specifies the scoring function to be used for ranking.
--best <int>
[OPTIONAL] With this option only a limited number of best scoring structures, as defined by --rankby, are reported in the three possible output files. If the --cutOff option is also specified, all best scoring structures are first passed through that filter. The user can specify the number of best scoring structures that should be reported.
--rankBy <TANIMOTO|TVERSKY_REF|TVERSKY_DB>
[OPTIONAL] This option defines the scoring used by the previous two options. More information about the three possible metrics can be found in the section on alignment scores. By default, the TANIMOTO measure is used.
-s, --scores <file>
[OPTIONAL] With this option a tab-delimited output text file can be generated, containing all results in a text-readable format.

The format of this optional scores output file (-s or --scores) is as follows:

column    Content
------    ---------------------------------------------------------------------
     1    Id of the reference structure
     2    Maximum volume of the reference structure
     3    Id of the database structure
     4    Maximum volume of the database structure
     5    Maximum volume overlap of the two structures
     6    Overlap between pharmacophore and exclusion spheres in the reference
     7    Corrected volume overlap between database pharmacophore and reference
     8    Number of pharmacophore points in the processed pharmacophore
     9    TANIMOTO score
    10    TVERSKY_REF score
    11    TVERSKY_DB score

Options

-f, --funcGroup <AROM|HDON|HACC|LIPO|CHARGE>
[OPTIONAL] By default the generated pharmacophores contain all functional groups and thus include all information that might be useful. With this option only a subset of the available functional groups can be used in the alignment. The user can define this subset by using the tags listed below with the ‘,’ symbol as separator. See below for examples. AROM: aromatic rings, HDON: hydrogen bond donors, HACC: hydrogen bond acceptors, LIPO: lipophilic spots, CHARGE: charge centers. If the reference and database structures are provided in the pharmacophore format then this option is discarded.
-e, --epsilon <double>
[OPTIONAL] This option can be used to change the tolerance for points to be matched in the alignment phase. This is an important parameter to control the feature-mapping phase as described before. The lower this value, the more strict the matching between two pharmacophores will have to be before they can be aligned. Higher values imply a higher allowed level of initial mismatching and typically result in larger computing times. The range of this parameter is between 0 and 1. The default value is 0.5.
-m, --merge
[OPTIONAL] Flag to indicate that pharmacophore points will be merged as explained above. Setting this flag also activates the -n or --noNormal flag because merged pharmacophore points do not contain a normal.
-n, --noNormal
[OPTIONAL] Flag to indicate that no normal information is included during the alignment. Using this flag makes the pharmacophore models less specific but also less conformation-dependent.
--noHybrid
[OPTIONAL] Flag to indicate that hybrid points should not be calculated. The list of hybrid pharmacophore points is given in the table above and is generated by default to reduce the number of pharmacophore points.
--scoreOnly
[OPTIONAL] Flag to indicate that the poses will be used as provided in the input file. No translational or rotational optimization will be performed. The best score reported is the one from the feasible mapping with the highest volume overlap.
--withExclusion
[OPTIONAL] Flag to indicate if the exclusion spheres should be part of the optimization procedure. By default, the overlap between pharmacophore and exclusion spheres is only taken into account at the end of the alignment procedure. When this flag is set, the exclusion spheres have also an impact on the optimization procedure.

Examples

In the first example the task is to generate pharmacophores for a number of structures and store these for later use:

> align-it --dbase db.sdf --pharmacophore output.phar

or shorter:

> align-it -d db.sdf -p output.phar

In the next example a virtual screening is performed. After screening a database against a reference structure, only the ranking based on the TANIMOTO score is of interest:

> align-it --reference      ref.phar
           --refType        PHAR
           --dbase          db.phar
           --dbType         PHAR
           --pharmacophore  output.phar
           --scores         result.tab
> sort -k 9 -r result.tab > sortedResult.tab

Calculating pharmacophores as in the first example, but without using hydrogen bond donor and acceptor information, is done by typing:

> align-it -d db.sdf -p output.phar --funcGroup AROM,LIPO,CHARGE

Finally an example is presented whereby a small fragment is used as a reference pharmacophore and the only purpose is to find structures that include this pharmacophore. Only structures with a common overlap covering at least 80% of the reference volume are reported in output.phar and result.tab. Notice that a ranking is made based on column 10 instead of column 9:

> align-it --reference      ref.sdf
           --dbase          db.phar
           --dbType         PHAR
           --pharmacophore  output.phar
           --scores         result.tab
           --cutOff         0.8
           --rankBy         TVERSKY_REF
> sort -k 10 -r result.tab > sortedResult.tab

Command line summary

Summary of the command line arguments to Align-it™:

GENERAL
-------
[O] -h, --help            N/A    Provides a short description of usage.
[O] --info                N/A    Provides a detailed description for each option.
[O] -q, --quiet           N/A    If this flag is set, minimum output is given to
                                 the user during execution of the program.
[O] -v, --version         N/A    Provides the version of the program.

INPUT
-----
[O] -r, --reference         -    Defines the reference molecule or pharmacophore
                                 that will be used to screen and/or align a
                                 database.
[O] --refType               -    Indicates the type of the reference data file.
[R] -d, --dbase             -    Defines the database that will be screened and/or
                                 aligned.
[O] --dbType                -    Indicates the type of the database data file.

OUTPUT
------
[R] -p, --pharmacophore     -    File with the computed pharmacophores of the input
                                 database.
[O] -o, --out               -    The transformed database molecules after aligning
                                 them to the reference pharmacophore.
[O] --outType               -    Indicates the type of the output molecular data file.
[O] -s, --scores            -    Tab-delimited text file with for each molecule the
                                 number of corresponding pharmacophore points and
                                 the overlap scores.
[O] -l, --log               -    Log file of the current run.
[O] --cutOff              0.0    Minimum score for a structure to be reported.
[O] --best                  0    Only best scoring molecules are reported.
[O] --rankBy         TANIMOTO    Define scoring used by --cutOff and --best.

OPTIONS
-------
[O] -f, --funcGroup       ALL    Flag to define functional groups used in the
                                 creation of pharmacophores.
[O] -e, --epsilon         0.5    Option to change the tolerance for points to be
                                 matched.
[O] -m, --merge           N/A    Flag to merge pharmacophore points.
[O] --noNormal            N/A    Flag to ignore normal information during alignment.
[O] --noHybrid            N/A    Flag to disable the use of hybrid pharmacophore
                                 points.
[O] --scoreOnly           N/A    Flag to indicate that the volume overlap should be
                                 computed from the given poses and that no
                                 translational or rotational optimization should be
                                 done.
[O] --withExclusion       N/A    Flag to add exclusion spheres into the optimization
                                 process instead of processing them afterwards.

Installation

Installation of the Align-it™ program relies on the libraries of OpenBabel version 2.3. Installation of OpenBabel is exemplified in the Configuring OS X for chemoinformatics section of this website.

The installation of Align-it™ assumes that the BABEL_DATADIR, BABEL_LIBDIR, and BABEL_INCLUDEDIR point to the directories where OpenBabel has been installed:

> echo $BABEL_INCLUDEDIR
/usr/local/openbabel/include/openbabel-2.0/
> echo $BABEL_LIBDIR
/usr/local/lib/openbabel/2.3.1/
> echo $BABEL_DATADIR
/usr/local/openbabel/share/openbabel/2.3.1/

Start by downloading Align-it™ from our software section and un-tar this file into your /usr/local/src directory:

> cd /usr/local/src
> sudo tar -xvf ~/Downloads/align-it-1.0.4.tar.gz

Change into this directory and start the building process:

> cd align-it-1.0.4
> sudo mkdir build
> cd build
> sudo cmake ..
> sudo make
> sudo make install

This latter command will install the Align-it™ executable in the /usr/local/bin/ directory. Finally, check the installation by entering:

    > which align-it
    /usr/local/bin/align-it
    > align-it -h
...

PyMol integration

PyMol is an open source visualization program well suited to produce high quality images of small molecules and biological macromolecules such as proteins. One of the main advantages of PyMol is its powerful scripting language.

The Align-it™ package can be downloaded from our software section. The package contains a align-it.py Python script to integrate the Align-it™ functionality into PyMol. The align-it.py script is located in the pymol/ folder of the downloadable distribution. The align-it.py script will be installed as a plug-in into PyMol.

Figure 7

Figure 7. Example of pharmacophore visualization in PyMol.

Installation on Mac OS X

In order to install the Align-it™ plug-in under Mac OS X, proceed according the following steps:

  • Change the application name MacPyMOL.app into Py-MOLX11Hybrid.app. You will probably find the MacPyMOL.app file in the /Applications directory:

    > cd /Applications
    > cp MacPyMOL.app PyMOLX11Hybrid.app
    
  • Copy align-it.py into the Py-MOLX11Hybrid.app directory (supposing that the align-it.py is located in the ~/Downloads directory):

    > cp ~/Downloads/align-it.py /Applications/PyMOLX11Hybrid.app/pymol/modules/pmg_tk/startup/
    
  • Start PyMOLX11Hybrid by double-clicking and activate the plug-in by choosing AlignIt from the Plugin menu. If successful, a new window containing a simple menu will show up.

Usage

The Align-it™ PyMol plug-in menu offers four options:

  • Create pharmacophores: to create a new pharmacophore the user first has to select a compound with the default selection name (sele). The Align-it™ tool should have been installed at /usr/local/bin/align-it (this should be the case if the installation procedure has been followed as described below). After executing Align-it™, all temporary files are removed and the pharmacophore is displayed and saved internally.
  • Read pharmacophores: instead of calculating a pharmacophore it is also possible to read a pharmacophore file. Only files with .phar extension are recognized. The pharmacophore is also saved internally.
  • Write pharmacophores: saves the last calculated or read pharmacophore in the file format as described above.
  • Create exclusion spheres: exclusion spheres are pharmacophore points and are generated based on the environment of the selected compound. All atoms of the target within a distance smaller than 4.5 Å of the ligand will correspond to a sphere with sigma 0.7 Å. The generated pharmacophore points are saved internally. If there was already a pharmacophore saved, they will be appended to it.

References

[1](1, 2) Green, J.; Kahn, S.; Savoi, H.; Sprague, P.; Teig, S. (1994) ‘Chemical function queries for 3D database search’, J. Chem. Inf. Comput. Sci., 34, 1297-1308 [acs/ci00022a012]
[2]Grant, J.A.; Gallardo, M.A.; Pickup, B.T. (1996) ‘A fast method of molecular shape comparison: a simple application of a Gaussian description of molecular shape’, J. Comp. Chem. 17, 1653-1666 [wiley/19961115]
[3]Taminau, J.; Thijs, G.; De Winter, H. (2008) ‘Pharao: pharmacophore alignment and optimization’, J. Mol. Graph. Model. 27, 161-169 [pubmed/18485770]

Revision history

Version 1.0.4

[released on July 12, 2013]

The columns in pharmacophore files are now allowed to be separated by whitespace (blanks and tabs) instead of only tabs.

Version 1.0.3

[released on August 30, 2012]

Updated the --refType and --dbType parameters so that the actual input file formats can be specified with these arguments as well (in prior versions, only the 'MOL' or 'PHAR' types where allowed, and in case that the 'MOL' type was specified, the actual file format had to be deduced from the file extension).

Added the --outType option to define the desired format of the output file with aligned molecules. When the --outType is not given, the format is deduced from the output file extension.

Version 1.0.2

[released on May 22, 2012]

Updated the FindOpenBabel2.cmake file (thanks to Abhik Seal) and corrected a bug in the PyMol interface so that a pharmacophore gets constructed only from the selected atoms (thanks to Emilie PiHan)

Version 1.0.1

Corrected some #infdef preprocessor rules that led to warnings on some compilers and included the definition of a NULL value (thanks to Anne Walter) [released on March 26, 2012]

Version 1.0.0

This is the first official release of Align-it™. The program is a successor of the program Pharao from Silicos and is branched out of version 3.0.3 of this program.

Additions to the original Pharao version include:

  • Porting the documentation to html and adding some improvements to the documents.
Free software foundation

Contents

Current projects

  • Optimisation of novel PPI's
  • Design of anti-viral compounds
  • Novel fragment-based descriptors

Follow us