Transcript RDConnect_Patrinos

electronic PharmacoGenomics Assistant (ePGA)
Alexandros Kanterakis1, Kleanthi Lakiotaki1, Evgenia Kartsaki1, Theodora Katsila2, George Potamias1, George P. Patrinos2
1Institute
of Computer Science, Foundation for Research & Technology-Hellas (FORTH), N. Plastira 100, Vassilika Vouton, 70013 Heraklion, Crete, Greece
2University of Patras, School of Health Sciences, Department of Pharmacy, University Campus, Rion, 26504, Patras, Greece
The electronic Pharmacogenomics Assistant (ePGA) has been envisaged as a web-based system (www.epga.gr) that offers two main
services to the engaged biomedical communities; (i) explore – a service to search and browse through established pharmacogenomic
gene-variant-drug-metabolizer status associations, and (ii) translate – a service to infer metabolizing phenotypes from individual
genotype profiles for all known pharmacogenes.
ePGA can be of benefit to health professionals, biomedical researchers, and the general public. In this context, a machine-learning
methodology (decision-tree induction) has been employed to induce generalized pharmacogenomic translation models from known
haplotype – tables that are able to infer the metabolizer status of individuals from their genotype profiles. Preliminary results are highly
predictive and indicate the potential of the whole approach as well as its impact in the clinic, even towards the use of a
pharmacogenomics card/electronic health record.
The Clinical/Research/Stakeholder Question
The ePGA tool
Personalization
Quality of information
Cost
Updatable
Current scientific Knowledge bases
×
√
√
×
Commercial Direct To Consumer (DTC) companies
√
?
×
×
ePGA
√
√
√
√
PGx Translation Service
PGx Explore Service
We developed an automated PGx translation algorithm, which infers metabolic phenotypes
from individual genetic (SNP) profiles. For each pharmacogene, and based on available
(PharmGKB) haplotype/allele tables, an individual’s genotype-profile is matched against the
available gene-alleles. Next, each inferred diplotype is assigned to a PGx phenotype status,
according to available “look up” tables [the algorithm was verified with the Affymetrix© DMET
Plus respective translation results]. The core of the translation process is implemented in the
open-source R environment (www.r-project.rg) and uses R Studio’s Shiny web-application
framework (http://shiny.rstudio.com/) to build its web-based interface.
The explore functionality couples drugs, genes and diplotypes with personalized
recommendations in a simple database schema. Currently, ePGA relies on the PharmGKB
dosing guidelines. Health professionals or biomedical researchers browse this information
through a lightweight web service. ePGA shows recommendations that form an expandable
tree with user friendly and printable content.
The novelty of ePGA rests in its ability to translate genotypes into PGx phenotypes and drug recommendations, based on state-of-the-art
PGx knowledge. ePGA acts as an “one stop shop” web portal.
ePGA has adequately (sample coverage>90%) assigned PGx phenotypes to 2,504 individuals for 33 out of 69 pharmacogenes
A statistically significant (p<10-10) phenotypic difference has become evident among 1000Genomes populations in 29 pharmacogenes
The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007-2013)
under grant agreement No. 305444 (RD-Connect)