Shape-based alignment of small molecules is a widely used approach in virtual screening, biostere replacement and also has multiple applications in analyzing potential off-target interactions and cross-reactivity of known drugs. Most existing software for aligning small molecules, both commercial and freely available, use the Tanimoto Coefficient (TC), which is a size-dependent scoring function for measuring molecular similarity. Moreover, the statistical significance of the molecular overlap is never reported. An efficient algorithm for aligning small molecules that provides a size-independent score for comparing molecules of different sizes and the statistical significance of the alignment, would greatly assist in finding new lead molecules.
We present a new computational method, LIGSIFT, for the structural alignment of small molecules. Compared with existing similar methods, our approach is novel in three respects: (1) a new size-independent scoring function for evaluating molecular similarity between small molecules is introduced, (2) a statistical assessment of alignment significance based on millions of random comparisons is provided, and (3) an efficient algorithm for large-scale applications is described and benchmarked on a standard database of active and decoy molecules for 40 pharmaceutically relevant protein targets, listed in the Directory of Useful Decoys (DUD). We expect LIGSIFT to have a significant impact in drug discovery studies.
References: Ambrish Roy and Jeffrey Skolnick, 2015, LIGSIFT: An open-source tool for ligand structural alignment and virtual screening.Bioinformatics (Oxford, England). 31(4):539-44. PDF