Integggrating CambridgeSoft ChemOffice Enterprise and...
Transcript of Integggrating CambridgeSoft ChemOffice Enterprise and...
Integrating CambridgeSoftg g gChemOffice Enterprise and TIBCO SpotfireTIBCO Spotfire
To make a best-in-breed Life Science data visualization and analysis platform
confidential
OverviewOverview
A quick review of the scientific data visualization and qanalysis problem.
CambridgeSoft informatics infrastructure as the data access component.
Spotfire as the data visualization and analysis component. Integrating to yield a best-in-class solution. Demo
confidential
ProblemProblem
Present drug discovery data to the scientists such that they can easily develop and compare hypotheses.
– Lots of data from multiple assays.– The key hypotheses change during the lifespan of a project.
View the problem through 2 lenses at Array:– Drug discovery and development– Translational Medicine
confidential
The structure of scientific dataThe structure of scientific data
Data visualization and analysis is not a new problem and is y pnot unique to life sciences.
BUT, there are some aspects of scientific data that render it more challenging than other fields.– Diversity of data types and end points.– Diversity of units.– Highly multivariate space.
Conceptually, the data space is a large, hierarchically organized sparse matrixorganized, sparse matrix.
confidential
Data shapeData shape
There are 2 traditional “shapes” to data:p– Tall and Skinny Compound Assay Result
AR123 MEK 20AR123 MEK 20AR123 BRAF 10AR123 AKT 5
– Short and Wide
etc.
Compound MEK BRAF AKTAR123 20 10 5
confidential
Shape of drug discovery data is hierarchicalShape of drug discovery data is hierarchical
Compound Compound average (e.g. Avg. IC50)
Batch Batch average (e.g.Average IC50)
(sample) Assay Run (e.g. IC50)
confidential
Shape of translational medicine data – 2 linked hierarchies
AE’sPK
ConmedsPD
DemogSubjectIHC Sample
Treatment“Omics”
Med HXInventory
confidential
The ideal solutionThe ideal solution
The ideal solution – Keeps track of the Compounds of interest or Subjects of
interest.– Allows the user to transition between levels of the data
hierarchy easily and elegantly.– Integrates form, grid (spreadsheet), and visual
representations of the data.Respects and employs the hierarchical nature of the data– Respects and employs the hierarchical nature of the data domain.
confidential
Why is data visualization important?Why is data visualization important?
Napoleon's march on Moscow 1812-1813 in a QSAR-like ptable:
Date Troop # Position Direction Temp11/01/1812 150302 120 12' 15", 37 23' 43" Attacking11/05/1812 132983 120 30' 15", 37 44' 13" Attacking11/12/1812 120302 121 02' 14", 38 45' 20" Attacking11/15/1812 115093 121 23' 41", 38 57' 21" Attacking…
01/23/1813 53203 121 23' 41", 38 57' 21" Retreating 001/31/1813 47392 121 02' 14" 38 45' 20" Retreating -1501/31/1813 47392 121 02 14 , 38 45 20 Retreating 1502/13/1813 32020 120 30' 15", 37 23' 43" Retreating -1002/17/1813 17203 120 12' 15", 37 23' 43" Retreating -18
confidential
…
So visual scanning important because of:So, visual scanning important because of:
Density of data.y Humans are (usually) visual creatures.
– Specifically, humans are good at spatial visual scanning vs.Specifically, humans are good at spatial visual scanning vs. temporal visual scanning.
– Thus, put the data side-by-side. Read Tufte’s books
– The Visual Display of Quantitative Information– Envisioning Information– Visual Explanations: Images and Quantities, Evidence and
NNarrative
confidential
OverviewOverview
A quick review of the scientific data visualization and qanalysis problem.
CambridgeSoft informatics infrastructure as the data access component.
Spotfire as the data visualization and analysis component. Integrating to yield a best-in-class solution. Demo
confidential
CambridgeSoft Infrastructure for Drug g gDiscovery data
BioAssay AssayADMETManual Feed BioAssay
Oracle DB
SubmissionWeb App
Manual FeedDesktop App
ADMETAutomated Feed
Data Mining
O l DB
Automated FeedServer App
Biology Data MartRegistration
Oracle DB
Oracle DB
PhysicalProperties
Oracle DB
PropertiesServer App
BioSAR ChemBioVIZ.NETChemDrawFor Excel Spotfire
confidential
Web App Desktop AppDesktop App Desktop and Web App
What we have today:What we have today:
All Efficacy, Bioavailability, and Toxicity data is fielded to a y, y, ycentralized Oracle data mart.
A meta-data database (BioSAR) and a variety of reporting tools sitting on this data mart and is used for data extraction– Research Assay History– ChemBioViz for Excel– BioSAR
confidential
Developing Translational Medicine infrastructureDeveloping Translational Medicine infrastructure
MedidataRAVE
SASfil
BioAssayWinNonLin
RAVEValidated eCRF
filesNightly output
SAS
BioAssay
PD data
WinNonLin
PK data processing
Translational Med.D t M t
Sample inventory
Clinical Data Mart
CSVfiles
Data Mart infoClinical Data Mart
ChemBioVIZ.NETChemDrawFor ExcelSpotfire
We’re not here yet, but most of the pieces are in place – we h t tit h th t th
confidentialDesktop App
For Excel
Desktop AppDesktop and Web App
have to stitch them together.
SpotfireSpotfire
Well-established best-in-breed general data visualization gand analysis tool.
Widely used in the pharmaceutical industry. Includes very robust data modeling features and a fantastic
API.
confidential
TIBCO Spotfire Enterprise Analytics –yPlatform Overview
Managers, Consumers,Executives
AnalystsIndependentsConfigurators
AnalystsInformaticiansStatisticians
Managers, Consumers,Executives Informaticians
Statisticians
CLIE SpotfireWeb Player
(*) SpotfireEnterprise Player
SpotfireProfessional Spotfire S+ Spotfire Miner E
NTS
(*) Advanced Computational EnginesZero install web client Visual, Analytic & Dynamic In-Memory Engines
Web Player Enterprise Player Professional
SpotfireD l
pp
SE
RV
ER
Developer
Spotfire Web Player Server Administration & Integration
Spotfire Server
(*) Spotfire Application Data
(*) Event Processing Computation Engine
(*) Spotfire Statistics Services
(*) Spotfire Automation Services
RS
IT / DBA
In-Memory EnginesApp Data Connectivity
Application Data Services
Real-Time Connectivity
Processing Services - OA
Computation Engine( ) Spotfire Automation Services
SDKs
DA
TAS
OU
RC
E RDBMS
Development,Solutions,Integration Flat
FilesWeb
ServicesSpread-sheets
SAP R/3SAP BW
OracleE-Business
SiebeleBusiness
Event Data StreamsCustom
confidential
ES
(*) = Optional components
OverviewOverview
A quick review of the scientific data visualization and qanalysis problem.
CambridgeSoft informatics infrastructure as the data access component.
Spotfire as the data visualization and analysis component. Integrating to yield a best-in-class solution. Demo
confidential
Timeline and approachTimeline and approach
In the spring and summer of 2010, we reviewed the state of p g ,the art in life science data visualization and analysis tools.
Nothing was a great fit for our needs and budget. Array approached CambridgeSoft and Spotfire and
proposed a new integrated solution .– Both the new ChemOffice Enterprise and the new Spotfire
DecisionSite are .NET applications.Th h b i t ti t th t bl– They each bring strong expertise to the table.
– Array has a long history of working with best-of-breed solution providers to create novel integrated solutionsproviders to create novel integrated solutions.
confidential
ProcessProcess
First proposal made in August/Sept. of 2010.p p g p First proof-of-concept completed at the end of October. The last ~5 months have focused on converting the proof-g p
of-concept into productized code. The full system is installed and operational on a
development infrastructure at Array We plan to go to production at the end of Q2.
confidential
Demo – example workflowDemo – example workflow
Focus on Drug Discovery example.g y p User:
1. Builds a form in ChemBioVIZ.Net.1. Builds a form in ChemBioVIZ.Net.2. Queries for Batches of interest.3. Pulls the data into Spotfire.p4. Identifies a key comparison of interest.5. Aggregates the key assay data up to the Batches level.gg g y y p6. Builds the key plot.7. Identifies the lead compounds to advance.
confidential
ChemBioVIZ Net - Form viewChemBioVIZ.Net - Form view
Any number of Forms, Tables, Assemble dataAny number of Forms, Tables, and query interfaces.
Assemble data from assays and aggregate to any level of the data hierarchyhierarchy
Queries permits pfor tracking lists
of compounds of interest and
list logic.g
Send to Spotfire
confidential
Send to Spotfire.
ChemBioVIZ Net – Dataview treeChemBioVIZ.Net – Dataview tree
Allows for administrative control of how the users access data.
Organized by project, therapeutic area, etc. as needed.
confidential
Clicking “Send to Spotfire” for its robust capabilitiesClicking Send to Spotfire for its robust capabilities
Example, enzymatic assay vs. cell assay, color by Rule of 5 violations.
Mouse-over to get details.
Select to populate
Details-on-demand showing t t
Select to populate drill-down chart.
structure
confidential
Spotfire – SAR tableSpotfire – SAR table
With rich qualitative and quantitative coloring (structures hidden in this examle).
confidential
From within the Spotfire workflow – modify ythe data in the analysis Results criteria editor permits application-independent authoring of the
data to view.
confidential
Results Criteria Editor featuresResults Criteria Editor features
Fast access to all tablesFast access to all tables in the dataview.
Fast form building.A t b t Aggregate between hierarchy levels.
Drag-and-drop column organization.
Quick filtering of available columns.co u s
Column renaming.
confidential
Features of the integrated solutionFeatures of the integrated solution.
ChemBioVIZ.Net and the Spotfire view are in sync. p ySelecting a point in Spotfire places that Compound in view in the Form viewer.
The Results Criteria will allow for a user to transition from ChemBioVIZ.Net to CBV for Excel to Spotfire easily.
The Spotfire analysis file “remembers” it’s Results Criteria The Spotfire analysis file “remembers” it’s Results Criteria and thus can be launched independently.
Loosely coupled but tightly integrated solutionLoosely coupled, but tightly integrated solution.
confidential
CambridgeSoft and Spotfire – current statusCambridgeSoft and Spotfire – current status
The proof-of-concept was very successful – quick p p y qimplementation and the integration works well.
Work is ongoing to productize the solution - Array plans to have a first implementation in production at the end of Q2.
Spotfire deployed at Array.– Training ongoing– Value being generated even with flat-file and SD file
i t tiintegrations. Combined solution promises to proved Drug Discovery and
Translational Medicine solutions to Array.Translational Medicine solutions to Array.
confidential