Text of Norwegian clinical genetics analysis platform ”genAP”, Thomas Grünfeld and Tony Håndstad, Oslo...
1. Norwegian clinical genetics analysis platform genAP VERDIKT
Conference October 15, 2013 T. Grnfeld og T. Hndstad
2. Agenda Drivers and development of individialized medicine
Our areas of focus, and some challenges Selected examples
3. Falling prices.
4. Faster and faster, more and more. Increasing number of
persons having their genome mapped
5. Possible development of full whole genome sequencing Sport
for the rich and famous The first scientists etc. Minisolution
23andme Offer for those Easier to Deployment in analyze all
screening who really needs it than 3-5 genes (E.g. newborns)
Patients with challengin diagnostics New causative genes Allows for
reuse of sequencing (DNA static) Later in-silico analysis
Simplifies existing methods Allows for reuse WGS integrated part of
EPR Everybody routinely analyzed Integrated expert systems in EPR
Fra Henry T. Greely, direktr Centre for Law and biosciences,
Stanford University
6. Some major challenges Do we want to know everything? Privacy
issues Insurance companies Legal DNA registries Data security How
do we know what to do?
7. Agenda Drivers and development of individialized medicine
Our areas of focus, and some challenges Selected examples
8. Aim of the project The aim of the project is to (contribute
to) develop an ICT infrastructure for central, secure storage of
human genome data, which allows for dissemminated use nationally
(And possibly internationally) Elucidate how to best analyze
sensitive genome data through existing (internet based) tools
Enable efficient (societally) deployment of genome data in
diagnostics and treatment Vision of becoming an ICT platform for
personalized medicine in Norway Key end products of the project
are: (Pilots for) practical tools for the general clinician at
their bedside activities An infrastructure and (organizatonal)
methodology for expansion of the pilots to other clinical domains
An ICT infrastructure that can use aggregated data in the
continuous development of the solutions.
9. Clinical vs. Research focus Research Cancer and New
mutations immunology(?) Quality above newest technology Robustness
above flexibility Diagnostics Known mutations Cancer; therapy
Germlinemutations Somatic mutations Our success is clinical demand!
Other legal limitations and framworks than research solutions User
interface key: I.e. refinement of info to easily available
knowledge/ decision support
10. Overview of modules in the system
11. Some key issues Quality and standardization of analyses How
to convey the information Storage, ownership and datasecurity
Involvement of relatives Research vs. Clinical diagnostics Are the
lab. analyses of adequate quality? Major interpretation challenges
Where is the limit for what should be conveyed (Significance,
certainty etc)? Who and how should info be conveyed (Degree of
counseling, preparation) How often must data be re-interpreted,
providing for new knowledge Dataconfidentiallity vs. Availability
for usage: a trade off? Patient autonomy vs. Documentation
requirements How to deal with tests that have direct implications
for ? What when the patient dies? Increasingly blurry boarders
between what is research (Common interest) and clinical diagnostics
(Individual interest) Strong need to revise and update legislation
and guidelines From Henry T. Greely, Head of Centre for Law and
Biosciences, Stanford University
12. A new field and language in medicine (Bass Hassan, Univ of
Oxford) Research Clinical testing Clinical treatment BIOMEDISH
Bioinformatics Medical informatics Informatics
13. Challenges for HTS data analysis (Bass Hassan, Univ of
Oxford) Challenge Details Rare skill which is difficult to find
Bioinformatics competence Costly resources when found Experience
invaluable Self teaching very challenging Often not intuitive
(Demanding user interface) Software Demand for many different
applications Requires often command line skills These are just
about our experiences! Substantial amounts of data to handle ICT
infrastructure Requirements for high performance computing (HPC)
Resources are difficult to find and challenging to share
Substantial cost of storage and maintenance
14. Agenda Drivers and development of individialized medicine
Our areas of focus, and some challenges Selected examples
15. Automating variant analysis Sample volume is expected to
grow rapidly. The task of analysing genetic variants constitute a
bottleneck in the whole process. To speed up and increase the
quality of the analysis, we seek to automate parts of it and
provide decision support for the molecular biologists that analyse
the variant data.
16. Algorithm for variant analysis
17. Evaluate frequency and inheritance DMG cancer workflow Add
note (+class 2) Possible carrier/ comp heteroz With sufficient
population size; not a patient pop YES Note: X-linked inheritance
is a third possibility for other cases (esp. for general genetics)
Extract var Heterozygous? YES/ NO Evaluate frequency: >~10%
(lower for genot) Recessive Inheritance mode? Latest dbSNP build
not in Alamut, checks web if variant not found Note: in addition to
inheritance, frequency cut-off should be based on prevalence data
and/or frequency of known pathogenic variants (adjusted for each
gene/diagnosis) Common? NO Pat DB dbSNP Dominant Homozygous?
(hemizygous?) YES Add note (+class 1) YES/ NO Evaluate frequency:
>1% (lower for genot) With sufficient population size; not a
patient pop YES Answer/add as class 1 Always check BIC Report
18. DEMO
19. Providing decision support for clinicians The average
physician has little knowledge of genomics. For genomics to change
clinical practice, the information must be translated into
actionable recommendations, easily available in the form of a
decision support system.
20. Decision support prototype for Tacrolimus dosage tested on
transplant surgeons
21. Publication
22. Acknowledgements Morten C. Eike, Ph.D. (post doc) Dag
Undlien, prof. M.D. Ph.D. (project owner) Halvard Lerum, Ph.D. (EHR
integration) Lars Retterstl, M.D. Ph.D (lab doctor) Tim Hughes, Ph.
D. (variant calling) Eidi Nafstad, MSc. (lab engineer)