The document summarizes Pfizer's use of the tranSMART platform for various genomics and clinical data analyses including genome-wide association studies (GWAS), supporting exploratory data types like metabolomics and FACS data, and large collaborative efforts like the Alzheimer's Disease Neuroimaging Initiative (ADNI) and Parkinson's Progression Markers Initiative (PPMI) datasets. It also discusses analytical integration with Genedata Expressionist and plans for future enhancements to tranSMART like improved GWAS support and additional genotype data. Contributors to these efforts are acknowledged.
See the 2,456 pharmacies on the National E-Pharmacy Platform
tranSMART Hackathon Paris - Nov 5-7 2013
1. tranSMART Hackathon
Paris – Nov 5-7 2013
tranSMART and Pfizer
GWAS, Exploratory Research and Preparing for Large Scale
Genotyping
Jay Bergeron
Director, Translational and
Bioinformatics
Pfizer inc.
Cambridge Mass
Co-Lead, Platform Development
eTRIKS
Oct 07-08, 2013, Berlin, Germany
2. Clinical/Translational Data Flow
CRO1
Downstream Systems
Medical
Safety
Regulatory
PharmSci
ClinPharm
Portfolio
Finance
Documents
…
Clinical
Trials
Management
CRO2
CRO …
1
Operational
reporting
Clinical data
analysis
Submission Use
2
Public data
Molecular data
from PFE studies
1
Exchange of clinical or molecular
data as needed
2
Feed specialty analysis systems
and retrieve results
Exploratory
Analysis
Specialty systems
Gene
Expression
Systems
Biology
Other
Exploratory Use
3. Pfizer’s Use of tranSMART
Translational Studies
Vaccines
Vaccines
Neuro
Neuro
PPMI
PPMI
Neuro
Neuro
TRACK
TRACK
Immuno
Immuno
Neuro
Neuro
ADNI
ADNI
Inf&Reg
Inf&Reg
NEXT AIBL?
NEXT
NEXT CAMD?
NEXT
CVMED
CVMED
Genotypes
Immuno
Immuno
Phase 33
Phase
2300
2300
BioBank
BioBank
Clinical
Clinical
Pipelines
Pipelines
Collection
Collection
CVMED
CVMED
Expression
Expression
More
NEXT
NEXT Genotypes
GWAVA
GWAVA
>150 Case Control
>150 Case Control
12 eQTL Loaded
12 eQTL Loaded
>300 Metabolomic QTL
>300 Metabolomic QTL
GWAS
BioTx
BioTx
>300 Case Control
>300 Case Control
PharmaTx
PharmaTx
Oncology
Oncology
Trials-Exploratory
7. Contribution: GWAS
GWAS search enabled in tranSMART using the Faceted Search
component with filtering by study, analysis and region of interest
8. GWAS Search Upload
Query Across GWAS Analyses
Search of Region of Interest
GWAS-Specific User Data Load Form
9. Interactive Manhattan Plots
GWAVA
Java Web Start
Thick Client
-Search Analyses by Genes/RS IDs
-Generate Manhattan Plots
Overlays and Trellises
Presentation Ready
-Generate Tabular Results
10. Access via the Bioservices
Platform (eQTL/mQTL)
A supplement to the tranSMART
user interface where necessary
11. GWAS
Data Verification and Correction
Correction
GWAS files can be mapped to specific field names
Checked for errors prior to loading
Records having errors can be removed
12. GWAS
In general we could argue that nearly all of the projects/targets that we review benefit from
tranSMART/GWAVA because the tools allow us to quickly review targets for a broad range of
phenotypic associations. In some cases we find additional associations, in some cases we do not.
The ability to do this quickly and efficiently is important.
- Geneticist, Precision Medicine
~500 GWAS Analyses to Date
12 EQTL, 1 MQTL
~700GB
GWAS: Second Development Effort
Plug-in Architecture
GWAS Filter Enhancements
Full integration of GWAVA
Document Representation
Saved Searches (Dataset Explorer)
Commitment to Open Source
Recombinant and the Hyve
Large Scale Queries
Considering MPPs
13. Genotypes
Genotypes: Access via the R-API and Bioservices
2300 Genotypes
From a
Phase-3 Study
Access via the R-Serve Interface
Currently Accessed via a Bio Service
10’s of thousands additional genotypes
expected in 2014
14. GWAS Contributors
Pfizer BT*
Christoph Brockel
Angela Gaudette
Peter Henstock
Ami Khandeshi
Michael Miller
Anna Silberberg
Kurt Watrous
Haiyan Zhang
Rohit Ranjan
*BT: Business Technologies
Pfizer RUs*
Eric Fauman
Janna Hutz
Scott Jelinsky
Katrina Loomis
Sara Paciga
Craig Hyde
Matt Pletcher
Nadeem Sarwar
Ciara Vangjeli
Gemma Wilk
Li Xi
*RU: Research Unit
Recombinant*
Dina Aronzon
Bob Coopersmith
John Gagnon
Devon Johnson
Jinlei Liu
Michael McDuffie
David Newton
Nancy Pickard
Raveen Sharma
Chris Urich
Haiping Xia
*Recombinant: Recombinant by Deloitte
19. Exploratory Molecular Data
Predominantly Low Dimensional Representation
Possibly dual (High and Low) representation
Clearly exposes data set explorer export limitations
Most Exploratory Data Loading Performed in House
Haiyan Zhang
All Pfizer ETL Engineering
Ami Khandeshi
21. Use case: ADNI and PPMI data in and PPMI in tranSMART
ADNI tranSMART
Enabling rapid exploratory analyses and hypothesis generation
Enabling rapid exploratory analyses and hypothesis generation
•
Data obtained from consortia and
collaborations are often poorly utilized and
have limited distribution across Pfizer
• Isolated, local storage of datasets
• Multiple, incomplete versions
• Duplication of efforts to transform data
•
Two large Neuroscience datasets were
chosen for addition to tranSMART
• Further evaluate the utility of tranSMART
for exploratory data analysis by
researchers
• Permit the development of processes for
data importation and handling
• Establish tranSMART training processes
22. Overview ADNI Datasets in tranSMART
8,000,000 Data Points
LONI Data Download
Approximately 140 excel
spreadsheets
2972 subjects
~8,000,000 data points
ADNI Dataset
23. Overview PPMI Datasets in tranSMART
~715,000 Data Points
LONI data
download
approximately 75
excel
Spreadsheets
780 subjects
~715,000 datapoints
PPMI Data Set
24. What were the ADAS-Cog11 scores at
Overview PPMI Datasets in tranSMART
screening?
Drag and drop Numerical Data from the Study
Navigation to Results/Analysis tab to Get t-test statistics
24
25. What tranSMART
Overview PPMI Datasets in are the baseline demographics of
Healthy and AD subjects in ADNI1?
NL
AD
26. What were the Base Assessment Scores
Overview PPMI Datasets inthe Healthy Control vs AD cohorts?
for tranSMART
NL
AD
27. Are There Differences in
Overview PPMI Datasets in tranSMART
CSF Biomarkers at Baseline?
NL
AD
28. TRACK-TBI
HDD Priority:
Impact of Consistent Systems
Limited use of clinical data obtained externally
One Mind for Research uses tranSMART
One Mind for Research uses tranSMART
Pfizer Obtained these Data Overnight Once the Agreement was In Place
Pfizer Obtained these Data Overnight Once the Agreement was In Place
29. Neuroscience Collaborative Studies
HDD Priority:Data Management and Collaborations
ADNI, PPMI, TRACK of clinical data obtained externally
Limited use
Availability of data in tranSMART allows exploratory analysis of the large datasets in minutes rather than
Availability of data in tranSMART allows exploratory analysis of the large datasets in minutes rather than
days or weeks – Director, Neuroinformatics
days or weeks – Director, Neuroinformatics
tranSMART: standard data
aggregator for Prec. Med.
ADNI , PPMI, TRACK
datasets imported
Initial training of end users
Follow-up will be performed
over the next 3 months
All new collaboration and
consortia proposals need to
include
Downstream data use, analysis
and management
Including budget/resources
30. Contributors
Pfizer BT*
Ami Khandeshi
Anna Silberberg
Haiyan Zhang
Robb Linde
Pfizer RUs*
Tom Comery
Jesse Macomber
Peter Bergethon
*BT: Business Technologies
Thomson Reuters
Sirimon Ocharoen
Ray Wright
IDBS
Mark Dekanter
Donnie Qi
Matt Clifford
Vladimir Kubatin
One Mind for Research
Srini K
Magali Haas
*RU: Research Unit
35. Contributors
Expressionist Integration
Pfizer BT*
*BT: Business Technologies
Genedata
Angela Gaudette
Peter Henstock
Andrew Hill
Ami Khandeshi
Anna Silberberg
Haiyan Zhang
Bill Mounts
Scott Jelinsky
*RU: Research Unit
Daniel Nesbit
Alice Li
James Cooper
Jens Hoefkens
Jessica Qi
Michael Riegelhaupt
Scott Faria
*Recombinant: Recombinant by Deloitte
36. HDD Priority:
Final Thoughts
Limited use of clinical data obtained externally
Plans
GWAS in 1.2
Genotypes
Analytical Integration
Collaborative
tranSMART Foundation
eTRIKS
others?
Outreach
Helping commercial entities find value in the
tranSMART community
37. Contributors
Pfizer BT*
Christoph Brockel
Angela Gaudette
Peter Henstock
Andrew Hill
Ami Khandeshi
David Klatte
Michael Miller
Anna Silberberg
Padma Reddy
Kurt Watrous
Haiyan Zhang
Anita Pracheta
Rohit Ranjan
Thomson Reuters
Sirimon Ocharoen
Ray Wright
*BT: Business Technologies
Pfizer RUs*
Eric Fauman
Scott Jelinsky
Katrina Loomis
Sara Paciga
Stephanie Hall
Craig Hyde
Nadeem Sarwar
Michael Swietek
Ciara Vangjeli
Li Xi
Tom Comery
Jesse Macomber
IDBS
Mark Dekanter
Donnie Qi
Matt Clifford
Vladimir Kubatin
*RU: Research Unit
Recombinant*
Dina Aronzon
Jinlei Liu
Michael McDuffie
David Newton
Nancy Pickard
Raveen Sharma
Haiping Xia
John Gagnon
Genedata
Daniel Nesbit
Alice Li
James Cooper
Jens Hoefkens
Jessica Qi
Michael
Riegelhaupt
Scott Faria
*Recombinant: Recombinant by Deloitte
Hinweis der Redaktion
This slide is an update of Tom Comery’s slide.
The support of these Neuroscience longitudinal studies is pivotal with respect to Pfizer’s use of tranSMART. Thousands of subjects and thousands of endpoints.
The cost for loading the ADNI (Alzheimers Disease Neuroimaging Initiative) set was $57K plus ~$5K for mapping with Meddra and WhoDrug standards. Loading took about 2 months. The vendor, Thomson Reuters, had prior experience with this data that I’m sure led to cost and time efficiencies.
PPMI (Parkinsons Progressive Markers Initiative) cost $52K and also took ~2 months. The vendor was IDBS. In this case, IDBS used a proprietary data mapping/transformation software that necessitates Pfizer to pay IDBS for updates (~$2K per update). If the ADNI file formats do not change, it is likely that we could load these data without Thomson Reuters if needed.
TBI is an (anti-climatic) but important example of what the Pharma’s are trying accomplish with tranSMART. One Mind for Research (Brokers of the TRACK-TBI) uses tranSMART. When an agreement was reached with Pfizer, One Mind provided the tranSMART data and mapping files to Pfizer (at ~5pm on a Tuesday) and the data was loaded by the next morning with only two errors associated with changes made to tranSMART by One Mind that were easily resolved. Cost to Pfizer for data-management, less than a couple of hours of an FTE. 0 budget cost.
These sets are only accessible by authorized personnel based on agreements with One Mind and the Laboratory of Neural Imaging at UCLA (PPMI, ADNI). tranSMART access privileges are set appropriately.