Chemistry Reference
In-Depth Information
of GRIND is that the molecular interaction fields can be regenerated from the autocorrelation transform and the
results of the analysis represented graphically together with the original molecular structures, in 3D plots.
ALMOND is a software package for the computation, analysis and interpretation of GRIND yielding highly
predictive and interpretable models [444-447]
For identifying regions of strong interactions between biological receptors and ligands, the MIF`s can be extremely
useful yielding important applications in drug discovery. Using analytic expressions, the MIF can be sampled at
regular intervals over the space surrounding the molecules at certain `grid modes` yielding thousands or hundreds of
thousands of data points for regular size molecules. Of particular interest are the `hot spots`, i.e regions of space
holding the most negative (favorable) energy values representing potential locations where a ligand can place a
functional group similar to the probe. These hot spots, which can be identified by visual inspection of graphical
representations, can also represent groups of the receptor binding site with which the molecule could establish
favorable binding interactions.
Hot spots can also be extracted by computational methods summarizing the most relevant information contained in
hundreds of thousands of MIF nodes. Although in principle, not simple, the development of computational
algorithms to extract these regions has potential applications in docking simulations, molecular superposition
algorithms and other areas. Various methods have been proposed for obtaining hot spots from MIF as well as from
grid MIF. Most of the methods proposed have drawbacks and different limitations that hamper their general
application in the field of drug design [444-449].
Among these methods, AMANDA [448] is a simple but efficient algorithm which can be used to extract a set of hot
spots from any MIF which can be applied to small molecule design. This algorithm requires that the starting MIF
nodes are tagged by the atom contributing most to the field energy. A node prefiltering is first carried out by
applying a energy cutoff in order to discriminate between relevant nodes and those representing weak or nonspecific
interactions, which are removed. A list of remaining m
i
nodes assigned to every atom (
i
) in the molecule is then
built. A nonlinear fraction of the total number of nodes is determined, containing a few nodes for small regions and
more nodes for larger regions.
First, the node with the lowest energy value is selected. Then the Euclidean distances between the chosen node and
the rest of the members of the list is computed and added to their field energy values to yield a simple scoring for
every mode and algorithm step. The node with the best scores is selected and the procedure is repeated. The
algorithm guarantees selection of at least one node for every atom of the molecule for which the list of prefiltered
nodes is not empty. The total number of nodes selected for a whole molecule depends on the number of atoms and
groups able to produce relevant interactions. In order to be useful for drug design, the hot spots must depict the
regions around a ligand which are more likely to participate in noncovalent bonding interactions with its receptor.
Consequently, the quality of a hot spot extraction algorithm depends on how complete and accurate this picture is. In
order to validate the method, hot spots for a collection of ligands were generated for comparison with experimental
data for a collection of ligand-receptor complexes [448].
There was recently reported [449] a high-throughput virtual screening of proteins using grid molecular interaction
fields. This new computational algorithm for protein binding sites characterization and comparison uses a common
reference framework of the projected ligand-space four-point pharmacophore fingerprints, includes cavit shape and
can be used with diverse proteins as no structural alignment is required. Protein binding sites are first described
using GRID molecular interaction fields (GRID-MIFs) and the FLAP (fingerprints for ligands and poteins) method
is then used to encode and compare this information. The discriminating power of the algorithm and its applicability
for large-scale protein analysis was validated by analyzing various scenarios: clustering of protein families in a
relevant manner, predicting ligand activity across related targets, and protein-protein virtual screening. In all cases,
the results showed the effectiveness of the GRID-FLAP method and its potential use in applications such as
identifying selectivity targets and tools/hits for new targets via identification of other proteins with
pharmacophorically similar binding sites [449-450].
EasyMIFs and SiteHound; a toolkit for the identification of ligand-binding sites in protein structures was recently
reported [451] whereas SiteHound uses molecular interaction fields (MIFs) produced by EasyMifs identify protein
