Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Debra Kassabian
Novartis Institutes for Biomedical Research (NIBR) IT
Moving away from ISIS software with ChemAxon's Help
ISIS/Draw
Draw Structures
Chemistry software at NIBR, before ChemAxon
| ChemAxon UGM September 2012 2
ISIS for Excel
Structures in spreadsheets
ISIS/Base
Personal structure databases
Personal reaction databases
CambridgeSoft
ELN
ChemOffice
Direct Cartridge
Chemistry search in Oracle
Duplication of functionality
ISIS software no longer
supported
ELN search very slow
Ceres (ISIS/Base + ISIS/Host)
Historical collection of Novartis reactions
Screening sample database
ACD
Other purchased databases
Chemistry software at NIBR
| ChemAxon UGM September 2012 3
CambridgeSoft
Chemdraw
CambridgeSoft
Chemdraw/Excel
ChemAxon
JChem Cartridge
Let’s ask ChemAxon
If they can help...
ISIS/Draw
Draw structures
Ceres (ISIS/Base + ISIS/Host)
Historical collection of Novartis reactions
Screening sample database
ACD
Other purchased databases
ISIS for Excel
Structures in spreadsheets
ISIS/Base
Personal structure databases
Personal reaction databases
Direct Cartridge
Chemistry search in Oracle
ISIS Software Usage at NIBR
| ChemAxon UGM September 2012 4
Emeryville Cambridge (HQ)
Horsham
Basel
Singapore (NITD)
Shanghai
Over 1600 scientists with ISIS software installed
Instant JChem Evaluation as ISIS/Base replacement
| ChemAxon UGM September 2012 5
A group of chemists in Basel, Horsham, Cambridge were given an introduction and asked to try it
Collected a list of use cases to test in IJC
Tried local reaction and structure databases
Set up IJC as interface to a large (10 Mi. structures) Oracle database running JChem Cartridge
Chemists liked it! IJC is a valid replacement for ISIS/Base
Fast data import (SDF, RDF)
Fast structure search
Form creation is easier than in ISIS/Base
Connecting to Oracle and searching worked well
A few missing
features were
also identified
IJC Evaluation - continued
Features requested:
Ability to export a subset of data from a central database into a local database
Better support to move projects from Development to Test to Production environments
Improved export to RDF for databases having several structure fields per record
Greater flexibility when adding data to an existing database
| ChemAxon UGM September 2012 6
Overall experience was very positive. Great support
and responsiveness from ChemAxon. Thanks
IJC Rollout – Deployment
| ChemAxon UGM September 2012 7
JChem Cartridge
SDF,RDF,CSV...
Local Database
Shared Database Project Team, Lab...
Single User
Java Webstart with UrlWebApp
Oracle instance running JChem Cartridge for shared projects
Customized security module for IJC to integrate with our authentication and authorization infrastructure
IJC Rollout – ISIS/Usage Survey
All ISIS users asked to fill out a survey on ISIS/Base Usage. Around 10% responded.
Some use only ISIS/Draw
Many local ISIS/Base databases
• Personal collections of structures
• Personal reaction databases
• Project databases
• Lab inventories
• Old biological data not available elsewhere
A few ISIS/Host databases we didn’t even know of...
Access ACD and internal reaction collections in Ceres
| ChemAxon UGM September 2012 8
ISIS/Base -> Instant JChem Migration
| ChemAxon UGM September 2012 9
What to do:
Export data from ISIS/Base as SDF/RDF
Choose between local and shared database
Set up a project in IJC
Import data
Create forms
How to do it:
Hands-on workshops with chemists bringing their
own databases
Step-by-step how-tos to print out
Short task-oriented videos
LDC Structures
ACD
MDDR, CMC
Metabolite
Chirbase
Current Chemical Reactions
Historical collection of Novartis reactions
Ceres – ISIS/Base + ISIS/Host
| ChemAxon UGM September 2012 10
New Web
application
IJC Rollout Summary
ISIS/Base -> IJC Migration is ongoing
Trying to use the possibility to store data in Oracle and create shared projects as much as possible
Scientists will be able to use ISIS Software until it breaks but we hope to persuade as many as possible to move
Greatest hurdle is not technology, but getting people to read their Email and come to the workshops!
| ChemAxon UGM September 2012 11
IJC beyond ISIS Replacement: accessing the NIBR data warehouse
| ChemAxon UGM September 2012 12
Data warehouse
Web Services
DART (Data
Analysis and
Reporting Tool)
Query Definition
+ Fields to
display
Data IJC Forms
Define Query
Define Fields to Display
View data in Spreadsheet
1
2
3
4
Query Definition
+ Fields to
display
Data
IJC beyond ISIS Replacement: accessing NIBR Chemical properties calculation Framework
| ChemAxon UGM September 2012 13
Web Services for
Calculation of
Chemical Properties
Currently not using ChemAxon’s calculators
Web-services framework to allow global deployment of chemistry backend tools
Called by Knime, Spotfire, Focus, Small Molecule Registration and other tools
Services being built and deployed by several groups within NIBR with example
services: Property calculators (e.g. Lipinski properties, pKa, molecular volume, etc.)
R-Group decomposition
Scaffold tree
name2struct
find activity cliffs
IJC integration with NIBR’s framework will provide consistency with values
displayed by other applications
SMILES/Compound Nr.
clogP, PSA....
Why build a new reaction-database application?
Current situation :
• Literature and historic (pre-eLN) reactions: stored in an ISIS/host database; searchable from ISIS/base.
• eLN reactions: stored in the eLN (CambridgeSoft) schema; “searchable” from eLN
Objective:
• allow NIBR scientists to quickly and efficiently search both sets of reactions in one place
Requirements:
• web-based UI for searching and browsing
• web services for searching and retrieving the data
| ChemAxon UGM September 2012 14
An Aside: services
Serious effort underway to expose every computation possible via web service.
Some examples:
• Querying our new data warehouse
• Calculating descriptors, using predictive models, etc.
• Registering compounds/biologics, querying registration system
Advantages:
• portability
• instantly global
• easy updates/deployments
• write once, use everywhere
• mashups | ChemAxon UGM September 2012 15
Reduction to practice
Collaboration with ChemAxon
Collection of user stories within NIBR
• global team of medicinal chemists driving requirements
• informed by Pistoia Alliance questions together with experience with CERES + various commercial tools
Concrete requirements and design done collaboratively
Implementation by ChemAxon
| ChemAxon UGM September 2012 16
Query Requirements vs. Data Challenges
| ChemAxon UGM September 2012 17
Solution: Bring over only the pieces of information that are actually needed for searching/browsing. Present additional data using an alternate “details” page that can be different for each original source
Standardize fields: Reaction, date, chemist, source, project, yield, solvent, reagents
Requires a custom RDF importer for each
historical data source
User wants to…..
Retrieve results based on structure searches (exact and substructure) of reactants and/or products
Retrieve additional information for a particular reaction
Query across multiple data sources
Limit results by yield, reagent, solvent, etc.
But we have data complexities…
Multiple collections of internal data in separate ISIS/Host databases with different structures
Constructing a single schema that can hold everything is unrealistic
Architecture
Oracle/JChem
Data Service
Layer (Tomcat)
Web application
(Tomcat/Wicket)
Web Browser
(javascript)
Oracle/eLN Update
application
Technical requirements:
Interface should be HTML5 and support a range of browsers
Must support our internal authentication/authorization framework
REST-style web services
eLN database updates must be automated
| ChemAxon UGM September 2012 18
Status
historic reaction data (~400K reactions) loaded
beta version of web interface and services complete
starting to test the application with our chemists
bug fixing
planning eLN -> JChem data migration (~600K additional reactions)
• Starting with lab notebooks for test users
• Obtain metrics for performance for export / import
• Plan incremental update
| ChemAxon UGM September 2012 19
Interface Overview
| ChemAxon UGM September 2012 20
Substructure search with results
| ChemAxon UGM September 2012 21
Search-results filters
| ChemAxon UGM September 2012 22
Additional search criteria + filters
| ChemAxon UGM September 2012 23
What’s next?
Get the application out to our test users and collect feedback
Get that eLN data loaded
Start thinking about better ways of organizing the results
Connecting to other web apps
Data analysis – unique reactions
| ChemAxon UGM September 2012 24
Acknowledgements
NIBR GDC/CPC
Ansgar Schuffenhauer
Graham Bloomfield
Nikolaus Stiefl
Peter Meier
Roger Fujimoto
Wolfgang Miltz
| ChemAxon UGM September 2012 25
NIBR IT
Anna Pelliccioli
Gregory Landrum
David Wayne Hughes
Manuel Schwarze
Rainer Stock
Sadik Keshwani
Stephanie Weeks
ChemAxon
Alex Drijver
Mihály Medzihradszky
Masoud Kalali
Tim Dudgeon
Thank you
| ChemAxon UGM September 2012 26