1 PowerMV Chemical Data Mining Environment S. Stanley Young Jun Feng and Jack Liu NISS MPDM, McMaster University 4 June 2005
2 Outline 1.PowerMV, a chemistry data mining environment. 2.rSVD, robust singular value decomposition, Liu, Hawkins, Ghosh,Young, PNAS (2003). 3.Space-filling designs. 4.PharmID: complex problem.
3 Environment 20k lines of interface code; 300k lines of algorithm code. ~0.75 man-years.
4 SD File, 50 years old and bad! 1 -ISIS- 3D V C N C C C C N C C S C C C C C C C C C C Etc.
5 Smiles Chemical Notation CC1=CC=C(C=C1)S(=O)(=O)N(CC(O)=O)C2=CC(Cl)=CC=C2 CC1CCC(CC1)C(C)(C)C(CC(C)C)C2CCCC(C)C2
6 Viewer
7 Molecule Blow Up
8 3D View
9 Compute Molecular Descriptors, Drug-Like Properties
10 Chemical Graphs and Properties
11 Multiple Descriptor Types Numerical Topology!
12 Similarity Searching, Motivating Paper
13 Select a Target Compound
14 Search Dialog
15 Structure Comparison Window Target Neighbor Annotation
16 Statistical Methods
17 Free download : Summary Display SD. Compute descriptors. Cluster. Statistical analysis. Similarity search. Become NISS affiliate! 20k display code 300k algorithm code
18 Contact Information Stan Young Jun Feng Become a NISS Affiliate!
19 Questions / Comments ?
20 Computed Properties – Rule of 5
21 Structure Comparison
22 Neighbor List
23 Attribute Comparison
24 Annotated Data Bases 1.Stockwell ACL 2.ChemBank 3.Exploratory Centers for Chemoinformatics Research 4.Chemical Entities of Biological Interest 5.Commercial, e.g. GVK, ACS, MDDR, Ashgate
25 Rule of Five
26 Example: CNS Compounds
27 Current NISS Research PharmID: Pharmacophore Identification Finds multiple binding modes! Selectivity. 1.NISS research. 2.Jun Feng, CompChem. 3.Seeking collaborators. 4.Become NISS affiliate.