novel methods for visual interpretation of biological screening data

Novel methods for visual interpretation of biological

screening data

Columbus Molecular Software, Inc.

30 March 2000

Overview of presentation

• Objectives• LeadScope Components

– Structural feature hierarchy– Structure analysis engine– Data visualization and dynamic querying– Statistics

• Applications• Future directions

Objectives

• Tool to interpret High Throughput Screening (HTS) results

• Analyze very large sets of structures and properties

• Accessible all scientist that wish to analyze HTS data– Computational chemists– Medicinal chemists– HTS

Chemical informatics for non-experts

• Easy-to-use GUI– “Window explorer” metaphore– Interactive sliders to query by properties

• Visualization using familiar terms and graphical objects

• Chemical normalization and analysis

Structural feature hierarchy

• Intellectually derived hierarchy of structural features

• Currently over 27,000 unique features

• Based on analyses of drugs

Query Structure for pyrrole, 3-amino(NH2)

and Matching Substructure

N

ONH2

O

O

N

NH2

atom is closed

Portion of the Pyridine Hierarchy

pyridine (level 1) pyridine, 1-R- (level 2) pyridine, 2-R- pyridine, 3-R- pyridine, 3-(alkenyl, acyc)- (level 3) pyridine, 3-(alkenyl, cyc)- pyridine, 3-alkoxy- pyridine, 3-(p-alkyl)- pyridine, 3-(s-alkyl)- pyridine, 3-(t-alkyl)- pyridine, 3-(alkyl, acyc)- pyridine, 3-alkylamino- pyridine, 3-alkylcarbonyl- pyridine, 3-(alkyl, cyc)- pyridine, 3-alkylthio-

Major structural classes

• Heterocycles

• Bases, nucleosides

• Naphthalenes

• Benzenes

• Natural products

• Carbocycles

• Peptidomimetics

• Carbohydrates

• Pharmacophores

• Elements

• Protective groups

• Functional groups

• Spacer groups

• User-defined features

Analysis of Dopamine

Total of 60 terms

NH2

OH

OH

Structure analysis engine

• Substructure searching incorporating– Aromaticity analysis– Tautomerism analysis– Generic group analysis– Check on atom environment

• No need to be expert in chemical conventions

Data visualization and dynamic querying

• Use of a variety of visualizations to represent sets of structures– Scatter plots, histograms

• Use of sliders to dynamically query dataset• Based around techniques developed at the

University of Maryland (Shneiderman and Ahlberg) †

†Shneiderman, Ben. (1998). Designing the User Interface: Strategies for Effective Human-Computer Interaction.3rd edition. Addison-Wesley.

Statistical analysis

• Uses statistical techniques to determine classes with unusually high numbers of active compounds

• Uses all the data (active and inactive)

• Color codes sets according to their correlation

Applications of LeadScope

1. HTS Data Analysis

2. Selectivity

3. Lead Optimization

4. Monomer Selection

5. Diversity Analysis

6. Structural Alerts

Demonstration

HTS Analysis

Monomer Selection

Diversity

Summary

• Chemoinformatic tool for non-experts– Easy-to-use interactive interface– Pre-defined analysis of chemical space– Behind-the-scenes chemically intelligent

analysis

• Tool encourages interactive participation

Future directions

• LeadScope Enterprise– Client-server– Access to entire corporate structure database– Links to property data in ORACLE

Acknowledgements

CMS

Paul Blower

Wayne Johnson

Julie Roberts

Kevin Cross

Glenn Myatt

Allen Richon

Pfizer

Mark Lord

Mike Snarey

Tony Woods

Paul Edwards