Transcript PharmGKB

Challenges Representing
Phenotype in Pharmacogenomics
Tina Hernandez-Boussard
PharmGKB
www.pharmGkB.org
Pharmacogenomics
 Understanding
how genetic
variation leads to variation in
responses to drugs
 A promise from the Genome
Project
 Personalized Medicine
– Making drug use effective and safe
based on a person’s specific
genotype
Pharmacogenomics Flow
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
PharmGKB: Capturing knowledge to
to catalyze pharmacogenomics research
PharmGKB Core Contents
Mission: aggregate, integrate & annotate
pharmacogenomic data and knowledge
PharmGKB Knowledge

VIPs
– Structured textual summaries of Very Important
Pharmacogenes and their key variants

Pathways
– Graphical pathway representations built by
consensus, associated with literature evidence
and links to PharmGKB genes, drugs,
phenotypes.

Literature Annotations
– PharmGKB curators create data entries that
associate genes with drugs and phenotypes,
based on an interpretation of the literature. They
encode with controlled vocabularies.
Genetic Variation Complexity

Genetic variation and its relation to proteins is
complicated
 “Gene” exists in the genome
 “Gene variations” specify the existence of
polymorphism:
– E.g. “There is A/C SNP at Golden Path X.”
– Haplotype variations = collection of simple variations

“Gene alleles” are specific variation options
– E.g. “One allele of the A/C SNP is A at GP X…”
– Haplotype alleles = collection of simple alleles
Genotypes are diploid alleles = “diplotypes”
 ASSOCIATIONS can be described to all of
these

Genotype-Phenotype Relations

Knowledge about gene-drug-pheno
interactions comes at different levels of
granularity:
1. Product of Gene X interacts with Drug Y (in pheno
Z)--in a physical sense
2. Variant of Gene X makes a difference in pheno Z for
Drug Y--in an association sense (can also be a
physical interaction, but that is with product)
3. Specific Allele of Variant of Gene X has a particular
effect on pheno Z for Drug Y--also in an association
sense
Mosaic Challenge: Throughput &
Redundancy
Limited curatorial staff has many duties
 Need methods to quickly identify
important knowledge and capture it in
computable form ONCE for multiple
uses
 With computable knowledge, can
generate displays appropriate for user
interests: pathways, VIP summaries,
literature summaries.

Goals for Representing
Knowledge in PharmGKB

Common platform for entering & curating
Pharmacogenomic knowledge = Protégé-based
– Pathways
– Very important pharmacogenes + variants
– Gene+variant-drug-phenotype associations

Structured entry for computability
– Standard vocabularies
– Automated linkages to existing data
• Genes, drugs, external resources
– Clear semantics

Extensible
– Usable SOON
– Expandable ALWAYS
Vocabularies Currently Used

HGNC for genes
– Gene families?

MEDDRA for adverse events
– Medical dictionary

MESH for disease, symptoms
– Vocabulary

Gene Ontology for cellular location, molecular
function, cellular biological process
 ASSUMES:
– Cell type vocabulary (MESH for now)
– chemical & drug vocabulary (MESH for now)
• Switch to chEBI for chemicals?
• Building drug dictionary @ PharmGKB
Knowledge Templates

Ingredients
– Controlled vocabulary of objects
– Logical representation of relationships
– Statement of key “slots” to be filled using objects,
according to logic.

EXAMPLE: Pathway Knowledge
– Pathway Overview template, points to “Steps”
– Pathway Step templates for
•
•
•
•
Metabolism step (PK)
Transport step (PK)
Inhibition step (PD!)
Downstream phenotype step (PK & PD)
Sample metabolism step
Sample Drug Interaction
Sample Phenotype Association
Conclusions
PharmGKB integrates, aggregates and
annotates data and knowledge to serve
the PGx research community
 Deep, high quality genotype data
 Phenotype data--mostly small studies,
some large ones in the pipeline.
 Knowledge services include literature
curations, pathways, VIP gene summaries
 Research efforts focus on creating pipeline
to improve efficiency and precision of
curated information

PharmGKB Team
Questions? Thanks.
[email protected]