Method for characterizing and analyzing 3-D shapes of molecules utilizing steric multiplets
View Patent ↗Steric features inherent in the three dimensional disposition of atoms in molecules can be represented as multiplets using a defined set of steric descriptors. The resulting multiplets can be encoded in a compressed form of bitstring known as a bitmap. Such bitmaps can be generated in compressed form and used to compare individual conformers or ensembles of conformers of molecules to each other without decompression. Such comparisons are useful in molecular similarity analysis, in molecular diversity analysis, in database searching, and in conformational analysis.
1. A computer implemented method for identifying the candidate molecule or molecules in a large array of molecules that are similar in three dimensional shape to query molecules, comprising the following steps:
a) defining a set of topologically distinct steric features that taken together cover the substructures found in each molecule;
b) identifying the positions of the steric features in each conformer of a query molecule;
c) selecting a class of multiplet for use;
d) using the selected multiplet class, for each query molecule, identifying the multiplet or multiplets that occur within at least one conformer of the molecule;
e) dividing the set of conformations for each query molecule into at least two subpopulations by randomly assigning each conformer to one subpopulation;
f) creating bitmaps indicating which multiplets occur in each conformer subpopulation of each molecule;
g) logically ORing the bitmaps of each conformer for each molecule;
h) generating a steric multiplet hypothesis by identifying discriminating multiplets from the bitmaps based upon how many conformers each multiplet occurs in and the size of the multiplet;
i) identifying the positions of the steric features in each conformer of the molecules in the large array;
j) using the previously selected multiplet class, for each molecule in the array identifying the multiplet or multiplets that occur within at least one conformer of the molecule;
k) creating bitmaps indicating which multiplets occur in each conformer of each molecule;
l) logically ORing the bitmaps of each conformer for each molecule; and
m) comparing the bitmap for each array molecule to the steric multiplet hypothesis bitmap to determine those array molecules most similar in three dimensional shape to the query molecule.
2. The method of claim 1 in which the selected multiplet class is chosen from either steric or pharmacophore-dependent multiplets.