VAMOS

Visualization of Accessible Molecular Space

A new compound filtering and selection interface

Spotfire User Conference - Europe - May 20 - 21, 2003

Jens Schamberger, Graffinity Pharmaceuticals AG

www.graffinity.com

About Graffinity

• Technology-driven drug discovery and pre-clinical development company

• Founded 1998

• Headquartered in HeidelbergGermany

• ~ 90 people

• Funding to date 41 M Euro

Technology

• Proprietary Screening Platform

• Chemical Microarrays

• Label-free Surface Plasmon Resonance Imaging

• Evolutionary Drug Discovery Strategy (RAISE®)

• Rapid Affinity Informed Structural Evolution

IntegratedDiscoveryPlatform

Chemical Microarrays Affinity FingerprintsData Mining

SPR Imaging

ParallelChemistry

Molecular ModelingMedicinal Chemistry

4608 Individual Compounds per SPR Image

Monomer 1

Mon

omer

2

Compound Acquisition

• Fragments forchemical microarrays

• Monomers forcombinatorial chemistry

• Reagentsfor lead optimization

Distributed Supplier Information

• Individual Vendors• Printed Catalogues• SD-Files• Databases

• Online Databases

• Compiled Collections• Available Chemical Directory (MDL)• Chemicals Available for Purchase (Accelrys)

• Virtual Collections

Need for supplier consolidation

Database Consistency & Integration

External DBCompanyInventory

?

Need for database merging

• External Databases• Individual Hviews• Limited Annotation

• Company Inventory• Well curated

• Virtual Databases

• Typical questions• Overlap• Similarity

Selection Criteria

• Range of Molecular Properties• Molweight, clogP• H-Don, H-Acc, Rot.Bonds

• Presence/ Absence of Functional Groups• Amines, Acids ...• MedChem ‘badgroups’ • Protecting Groups

• Presence/ Absence of Substructures/Pharmacophores• privileged structures• known scaffolds

Need for compound annotation

N

N

OO

ClHOOC

NO2

Filtering Procedures

• Hard Filters

• Individual Selection Criteria e.g. :• e.g. “at least one primary amine• and molecular weight 100 - 300 Da• and clogP smaller 5• and number of halogens smaller 4

• Even Lipinski criteria often taken as hard filter while it was meant as a rule

• No step back using hard filters

DATABASE

Filter

SubsetList

Need for flexible filtering tools

Rules and Profiles

Need for dynamic selection tools

• Soft Description withCase by Case Decisions

• Rules• Allowance for limited violation of criteria

• e.g. “no more than one Lipinski violation”• Definition of own rules

• Profiles• Subsets based on defined property profiles• Compound similarity based on common profile

Flexible filtering tools

Dynamic selection tools

Compound Selection from Databases

Supplier consolidation

Database merging

Compound annotation

Typical „spotfire what-iffing“

VAMOS

Perl

VAMOS Overview

MDL IsisDatabases

OracleN

NO

O

Cl

NO2

HOOC

NN

O

O

Cl

NO2

HOOC

Modelling andCheminformatics Data Storage

VAMOS: Merging of Database

ACD

NBB

ESC

IRC

AAvailable

CChemicals

DDirectory

NNew

BBuilding

BBlocks

EExternal

SScreening

CCompounds

IIn-house

RRegistered

CChemicals

ISIS Databases

Graffinity Unified Chemical DB

MDL Reagent Selector

• Isis DB with > 600,000 unique cmpds• complete source tracking• easily expandable by additional databases

TRIPOSUnity DB

VAMOS: Linking with Cheminformatics

OracleVAMOS

DB

Modelling &CheminformaticTools

ExtractMolecular

Data

Perl DBI

Graffinity Unified Chemical

DBStore

MolecularAnnotations

Standard Properties

• Identifiers• ACD number, NewBB ID, ChemID,...

• Supplier Information• Ranking, „Best Supplier“, ...

• „Lipinski-Like“ Properties• MW, ClogP, Don, Acc, Rings, Nrot,...

• Rule Based Features• Andrews Energy, ...

• Presence of Functional Groups• COOH, -NH2, -NH-, ......

Customized Properties

• Presence of Special Substructures• privileged fragments, bad groups, ....

• Cell Based Similarity • BCUT-Space, Substructure Classes, SOM

• Affinity Information• „Hit on Array“

• easily expandible

ISIS Database+ Oracle DB / Tab-Separated-Value File

________________________________= Customizable Molecular Spreadsheet

„one row one structure“

„one column one property“

NN

O

O

Cl

NO2

HOOC

NN

O

O

Cl

NO2

HOOC

VAMOS: Storing Molecular Annotations

VAMOS: Flowchart

ACD

NBB

ESC

IRC

GUCD

Pharmacophore and Substructure Query Definition (XML)

Create GuideHTML

OracleVAMOS

DB

Create/update

Initiate queriesGet query results

Store query results

SGI

OracleServer

New

New QueriesN

Cron Job

TRIPOSUnity DB

VAMOS: Spotfire Guide

ACD

CollectedCatalogues

In - house : colored

Source Database

goodvery good

reasonable unreliable

Supplier- Ranking

Similarity / Self Organizing Map

Donor - Acceptor Distribution

Profile Plot (Rule of Five)

Conclusions

• Integrated view across databases

• Extensive compound annotations

• Dynamic selection & real-time filtering capabilities

• Adjustable to specific compound acquisition needs

• Expandable for profiling of virtual libraries

• Adaptable for analysis of screening data

VAMOS : Visualization of Accessible Molecular Space

Acknowledgement :

Dr. Günther Metz

Dr. Dirk Tomandl

Helmut Wittneben

Graffinity Pharmaceuticals AG