Limits...
Implementation of the Hungarian algorithm to account for ligand symmetry and similarity in structure-based design.

Allen WJ, Rizzo RC - J Chem Inf Model (2014)

Bottom Line: False negative docking outcomes for highly symmetric molecules are a barrier to the accurate evaluation of docking programs, scoring functions, and protocols.For some families of protein systems the results are even more dramatic, with success rate increases up to 16.7%.Several additional applications of the method are also presented including as a pairwise similarity metric to compare molecules during de novo design, as a scoring function to rank-order virtual screening results, and for the analysis of trajectories from molecular dynamics simulation.

View Article: PubMed Central - PubMed

Affiliation: Department of Applied Mathematics & Statistics, Stony Brook University , Stony Brook, New York 11794, United States.

ABSTRACT
False negative docking outcomes for highly symmetric molecules are a barrier to the accurate evaluation of docking programs, scoring functions, and protocols. This work describes an implementation of a symmetry-corrected root-mean-square deviation (RMSD) method into the program DOCK based on the Hungarian algorithm for solving the minimum assignment problem, which dynamically assigns atom correspondence in molecules with symmetry. The algorithm adds only a trivial amount of computation time to the RMSD calculations and is shown to increase the reported overall docking success rate by approximately 5% when tested over 1043 receptor-ligand systems. For some families of protein systems the results are even more dramatic, with success rate increases up to 16.7%. Several additional applications of the method are also presented including as a pairwise similarity metric to compare molecules during de novo design, as a scoring function to rank-order virtual screening results, and for the analysis of trajectories from molecular dynamics simulation. The new method, including source code, is available to registered users of DOCK6 ( http://dock.compbio.ucsf.edu ).

Show MeSH

Related in: MedlinePlus

(a)ROC curves from docking 475 actives and 15 990 decoysto EGFR, then ranking the results by one of four metrics. The diagonal,indicated by a dashed line, represents random enrichment. (b, c) Crystallographicpose of reference ligand erlotinib shown as gray sticks and transparentsurface, identical in both panels, relative to the top four ligandsfrom the (b) combined value, shown as orange sticks, or the top fourligands from the (c) DOCK grid score, shown as orange sticks.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3958141&req=5

fig10: (a)ROC curves from docking 475 actives and 15 990 decoysto EGFR, then ranking the results by one of four metrics. The diagonal,indicated by a dashed line, represents random enrichment. (b, c) Crystallographicpose of reference ligand erlotinib shown as gray sticks and transparentsurface, identical in both panels, relative to the top four ligandsfrom the (b) combined value, shown as orange sticks, or the top fourligands from the (c) DOCK grid score, shown as orange sticks.

Mentions: For the rankingmethoddescribed in eq 5, #-ref atoms is the numberof non-hydrogen atoms in the reference ligand, and C1 and C2 are constants of−5 and 1, respectively, chosen such that the two terms areweighted approximately equally, and that greater negative values representmore favorable scores. Figure 10 presents thereceiver operating characteristic (ROC) curves derived from each ofthe four ranked lists along with structural overlays for top-scoringcompounds from two of the four methods. While visualization of theROC curves can provide a qualitative assessment of enrichment, resultsare also quantified below using area under the curve (AUC) computedacross the entire database (overall enrichment), and the number ofactives recovered after screening 1% of the database (early enrichment).


Implementation of the Hungarian algorithm to account for ligand symmetry and similarity in structure-based design.

Allen WJ, Rizzo RC - J Chem Inf Model (2014)

(a)ROC curves from docking 475 actives and 15 990 decoysto EGFR, then ranking the results by one of four metrics. The diagonal,indicated by a dashed line, represents random enrichment. (b, c) Crystallographicpose of reference ligand erlotinib shown as gray sticks and transparentsurface, identical in both panels, relative to the top four ligandsfrom the (b) combined value, shown as orange sticks, or the top fourligands from the (c) DOCK grid score, shown as orange sticks.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3958141&req=5

fig10: (a)ROC curves from docking 475 actives and 15 990 decoysto EGFR, then ranking the results by one of four metrics. The diagonal,indicated by a dashed line, represents random enrichment. (b, c) Crystallographicpose of reference ligand erlotinib shown as gray sticks and transparentsurface, identical in both panels, relative to the top four ligandsfrom the (b) combined value, shown as orange sticks, or the top fourligands from the (c) DOCK grid score, shown as orange sticks.
Mentions: For the rankingmethoddescribed in eq 5, #-ref atoms is the numberof non-hydrogen atoms in the reference ligand, and C1 and C2 are constants of−5 and 1, respectively, chosen such that the two terms areweighted approximately equally, and that greater negative values representmore favorable scores. Figure 10 presents thereceiver operating characteristic (ROC) curves derived from each ofthe four ranked lists along with structural overlays for top-scoringcompounds from two of the four methods. While visualization of theROC curves can provide a qualitative assessment of enrichment, resultsare also quantified below using area under the curve (AUC) computedacross the entire database (overall enrichment), and the number ofactives recovered after screening 1% of the database (early enrichment).

Bottom Line: False negative docking outcomes for highly symmetric molecules are a barrier to the accurate evaluation of docking programs, scoring functions, and protocols.For some families of protein systems the results are even more dramatic, with success rate increases up to 16.7%.Several additional applications of the method are also presented including as a pairwise similarity metric to compare molecules during de novo design, as a scoring function to rank-order virtual screening results, and for the analysis of trajectories from molecular dynamics simulation.

View Article: PubMed Central - PubMed

Affiliation: Department of Applied Mathematics & Statistics, Stony Brook University , Stony Brook, New York 11794, United States.

ABSTRACT
False negative docking outcomes for highly symmetric molecules are a barrier to the accurate evaluation of docking programs, scoring functions, and protocols. This work describes an implementation of a symmetry-corrected root-mean-square deviation (RMSD) method into the program DOCK based on the Hungarian algorithm for solving the minimum assignment problem, which dynamically assigns atom correspondence in molecules with symmetry. The algorithm adds only a trivial amount of computation time to the RMSD calculations and is shown to increase the reported overall docking success rate by approximately 5% when tested over 1043 receptor-ligand systems. For some families of protein systems the results are even more dramatic, with success rate increases up to 16.7%. Several additional applications of the method are also presented including as a pairwise similarity metric to compare molecules during de novo design, as a scoring function to rank-order virtual screening results, and for the analysis of trajectories from molecular dynamics simulation. The new method, including source code, is available to registered users of DOCK6 ( http://dock.compbio.ucsf.edu ).

Show MeSH
Related in: MedlinePlus