課件案例組會dicks report

1、On the Perception of Molecules from 3D Atomic CoordinatesReporter: Qian ZhangInstructor:Tingjun HouTutorialINTRODUCTIONMETHODRESULTS AND DISCUSSIONCONCLUSIONIntroductionRapid growth and large size of CSD,PDB databases have made them important resources in both the life and materials sciencesPDB stru

2、ctures are of poorer quality than the CSD, with lower resolution, missing hydrogen atoms, larger coordinate errorsThe ad hoc nature of the previous methods often limits their utility because they have been tuned for use with the PDB. Method: Determine connectionParamaters: denotes 3D coordinates of

3、n atomatomic number of n atomDetermine connection:is the “covalent radius” of atom iMethod: Determine connectionIt often happens that spurious bonds arise from coordinate errors(especially in the PDB) or strained geometries; such bonds require removal.For each atom i, a “dimension” is calculated as

4、followsFirst:Second:di=k, if k2di is the number of positive eigenvalues of A with square root greater than 0.2di=0, if the atom is isolateddi=1, terminal and linear atoms with at least two bondsdi=2, planar atomsdi=3, othersMethod: Determine connectionThird: Each pair (di,zi) determine an upper boun

5、d, BiFor each atom, only the shortest Bi bonds are retained from the candidate list. This removal may (rarely) result in an asymmetric connection table; i.e., atom i is in atom js list but not vice versa. Such asymmetric bonds are also removed resulting in a final, symmetric connection table.Method:

6、 Determine hybridizationdenotes the number of connections of atom ieach step is after application of the previous stepatoms with unassigned hybridizations have d 3,Z=C,N,O,Si,P,S,Se , Q 4 and at least one bonded neighbor with an unassigned hybridization. Method: Determine bonds orderAll bond ordersi

7、n which either atom i or j has non-“?” hybridization are set to 1For each bond (i,j ) not involving linear atoms the smallest out-of-plane dihedral is computed according to follows:If the computed smallest dihedral is greater than 15 degrees, then bij is set to 1Method: Determine bonds orderThe bond

8、 order is set to 1 ifderived from MMFF94 parametersMethod: Determine bonds orderAfter the bond length and dihedral angle tests have been applied, thehybridization of all uncharacterized atoms not involved in a bond of unknownorder sre set to sp3 Maximum Weighted Matching algorithm are applied then.a

9、ssign a weight Wij to each unassigned bond order bijapply the Maximum Weighted Matching algorithm to determine a matching with largest weightassign double or triple bonds to the edges of the match.Method: Determine weightthe edge weights are set to a sum of atom weights and bond length properties ac

10、cording to the following formulathe probabilities are estimated by counting frequencies of single and double bonding in a collection of 200000+ organic compoundsMethod: Determine weightElements with atomic number above 10 are mapped to second row elements in the same column of the periodic table and

11、 0.1 is subtracted from the weight in the table. A weight of -20.0 is used for elements not in the table. Method: Determine bonds order1.Once the maximum weighted matching is calculated, a bond length test is used to determine which of the match bonds should marked as triple bonds, this assignment i

12、s made only for linear atoms2.Two double bonds are assigned to a linear atom only if it and its neighbors are not matched.Ionization states and formal charges are perceived after the connectivity and bond orders have been determined.Results and discussionTestH-existOptimizedAccurate rateSourcecommon

13、 functional groupsNoYes100%common functional groups ring structuresNoYes100% biologicallyrelevant rings17 PDBsNoNo82%Ricketts122 PDBsNoNo93%Hendlich177 PDBsNoNo94% commonly used for the training of dockingResults and discussionPerhaps a softer criterion on bond lengths (instead of the hard cutoffs)

14、would solve this problem. ConclusionAdvantages:1.No ad hoc chemical group patterns nor ring analyses are used2.The method performed well on a collection of common functional groups and ring structures as well as on examples from the CSD and the PDB.Limitations:1.delocalized dative bond complexes are not handled well2.specific tautomeric form