The Discovery Informatics Framework
-
Upload
emerson-sloan -
Category
Documents
-
view
35 -
download
1
description
Transcript of The Discovery Informatics Framework
1
The Discovery Informatics Framework
Pat RougeauPresident and CEO
MDL Information Systems, Inc.
Delivering the Integration Promise
American Chemical Society Meeting
San Francisco, CAMarch 27, 2000
2
Integrating informatics into the Discovery process
TargetsTargets
Inventory
Proposals
X XX
Standard Test Set
X X XProof
Candidates
Descriptors (chem., physicochem. etc.)Met
ho
do
log
y (
alg
or.
)
Early Validation
safe new effectiveeconomical
LeadLead
Synthesis
RepeatAnd Repeat
3
DB
DB
Information sources for the Discovery process
Journals
Journals
Journals
Standard Test Set
TargetsTargets
Inventory
Proposals
X XX
X X XProof
Candidates
Descriptors (chem., physicochem. etc.)Met
ho
do
log
y (
alg
or.
)
LeadLeadSynthesis
Early Validation
safe new effectiveeconomical
DB
DB
DBJournals
4
Prioprietary information is exploding
High Throughput Screening
Combinatorial Chemistry
Genomics
Partnerships and Outsourcing
Mergers
5
Public information is more accessible
Globalized research
Globalized publishing
Electronic media
World Wide Web
Patent literature
6
Turn data into information assets
IT infrastructure
Information
ApplicationDrive
out cost
Drive up capability
InnovateEducate
GlobalizeIntegrate
StandardizeReduce costs
7
Turn information assets into actionable decisions & knowledge
Provide workflow tools that help ensure quality data
Provide access tools that give the right data at the right time
Provide analysis tools that help turn information into action
Capture the knowledge derived from this process for future use
8
Workflow tools: Assay Explorer
9
Access Tools
AR1
OH
OH
10
Analysis Tools
Humans are the best decision makers
Informatics must Aid the human ability to recognize
patterns through easy to manipulate visualizations of data
Improve UI’s to be more natural
11
Spotfire
12
Going beyond analysis to decision support
A truly effective decision support environment is build on an open informatics framework to Access all of the information available, in context Visualize and analyze against all or subsets of
the information Access tools for calculating and predicting
properties and predicting properties based on existing data
13
Going beyond analysis to decision support
Discover in silica predictive models
Test those models against existing data
Validate those models through additional screening
Result: Provide new leads more quickly, with fewer discovery cycles
14
Interoperating informatics solutions for Discovery
TargetsTargets
Inventory
Proposals
X XX
Standard Test Set
X X XProof
Candidates
Descriptors (chem., physicochem. etc.)Met
ho
do
log
y (
alg
or.
)
Early Validation
safe new effectiveeconomical
LeadLead
Analysis
CL ToolsCentral Lib
SMARTReagent Selector
CompoundWarehouse
CompoundWarehouse
ToxicityEcoPharm
Visualization
Assay Explorer
CompoundSelection
15
Accessing disparate data sources
BeilsteinDB
MDLDBs
EnterpriseDB
3rd PartyDB’s
ProjectDB
Compound Warehouse
Beilstein’s Application
MDL’s Application
YourApplication
YourApplication
3rd PartyApplication
16
Provide access to data anywhere: Compound Warehouse and LitLink
Beilstein MDL Enterprise 3rd PartyProject 3rd Party
Native Application
One query access to multiple databases
Compound Warehouse
LitLink Server
One click access from multiple databases
17
Facilitating interoperability
Decision SupportDecision Support Database BrowserDatabase Browser
Procurement Procurement
CWCW ResultResult Drill downDrill downQueryQuery
18
ContentContent TechnologyTechnology
Interoperability requires software and database resources
DecisionSupport
Your Application
Compound Locator
Database Browser ProcurementExperimental
Workflow
Knowledge Extraction
20
Knowledge—what scientists create
Recognizing and generalize patterns
Differentiating causality from coincidence
Recording conclusions in papers and reports, supported by data
21
Knowledge capture is key
In Discovery, capturing knowledge means capturing Decisions Analysis methodology Supporting data Context (e.g., experimental protocol)
22
Knowledge mining today
Today’s technology can help the scientist Search disparate sources Review the results Navigate between the sources
Recreate the knowledge
23
Knowledge extraction progress is being made
Automating knowledge base creation Intelligent indexing Automatic thesaurus construction
Mining the knowledge base Relevance based retrieval Natural language searching
24
Creative Science on a Systems Engineering Framework
Creative science is ad hoc interactive intuitive
Systems engineering is disciplined ordered structural
25
Creative Science on a Systems Engineering Framework
Change is a constant
Transitions require management
Take into account strategy pace values culture
26
Link business and scientific concerns
Science Business
People
27
Thank You