BCL::MolAlign: Three-Dimensional Small Molecule Alignment for Pharmacophore Mapping

Abstract

Small molecule flexible alignment is a critical component of both ligand- and structure-based methods in computer-aided drug discovery. Despite its importance, the availability of high-quality flexible alignment software packages is limited. Here, we present BCL::MolAlign, a freely available property-based molecular alignment program. BCL::MolAlign accommodates ligand flexibility through a combination of pregenerated conformers and on-the-fly bond rotation. BCL::MolAlign converges on alignment poses by sampling the relative orientations of mutually matching atom pairs between molecules through Monte Carlo Metropolis sampling. Across six diverse ligand data sets, BCL::MolAlign flexible alignment outperforms MOE, ROCS, and FLEXS in recovering native ligand binding poses. Moreover, the BCL::MolAlign alignment score is more predictive of ligand activity than maximum common substructure similarity across 10 data sets. Finally, on a recently published benchmark set of 20 high quality congeneric ligand-protein complexes, BCL::MolAlign is able to recover a larger fraction of native binding poses than maximum common substructure-based alignment and RosettaLigand. BCL::MolAlign can be obtained as part of the Biology and Chemistry Library (BCL) software package freely with an academic license or can be accessed via Web server at http://meilerlab.org/index.php/servers/molalign .

Figure 1.

Outline of the BCL::MolAlign flexible alignment algorithm. Rigid alignment is equivalent to a single tier of MCM optimization with a single conformation each for molecule A and molecule b. MC moves alter the current molecule A or B during each optimization tier. The same moves are used in each tier, but the number of steps differs in each tier (see the Methods).

Figure 2.

Schematic of sampling strategies implemented in BCL::MolAlign. From a given starting alignment on the left side of the arrow, the resulting alignment following each operation is depicted on the right side of the arrow. Once atoms and bonds have been chosen, BondAlign (A), BondAlign2 (B), and MatchAtomNeighbors (C) each have one possible outcome. BondSwap (D) has an equal probability of sampling two possible outcomes. Highlighted segments correspond to the chosen atoms and bonds for alignment. Atom numberings in MatchNeighborAtoms correspond to mutually matched pairs between molecules A and B.

Figure 3.

Rigid alignment of P38 inhibitors from PDB IDs 1OUK and 1OUY illustrates atom pairing at variable maximum atom distances. The 2D representations of the 1OUK and 1OUY ligands. The 3D representations depict the native pose of 1OUK rigidly aligned to the native pose of 1OUY. Spheres illustrate heavy atoms separated from a heavy atom in the opposite molecule by less than the specified maximum atom distance (D_max). Sphere radii correspond to half of the indicated maximum atom distance. Red and white overlapping spheres are considered matched atoms.

Figure 4.

RMSD to native binding pose as a function of maximum common substructure similarity across the AstraZeneca overlay sets. Comparisons were made for all molecule pairs (A), the best alignment for each molecule as measured by symmetry RMSD to the native pose (B), and for all pairs with a symmetry RMSD to native pose ≤2.0 Å (C). The substructures were defined by comparing the atom by element IDs and the bonds by bond order (including specification of aromaticity and/or inclusion in a ring). Similarity between each molecule pair is the maximum common substructure Tanimoto similarity.

Figure 5.

Visual representations of docked versus aligned poses in challenging docking targets. Comparisons show the protein–ligand complexes of the crystallized scaffold (gray) and crystallized target (white) molecules (A). The crystallized pose of the target molecule (white) is also shown with the RosettaLigand docked pose (green; B) and the BCL::MolAlign flexibly aligned pose (purple; C). Examples correspond to molecules from the HCV (row one), TPPHO (row two), and CTAP (row three) data sets.