ePGA: A Web-Based Information System for Translational Pharmacogenomics

Abstract

One of the challenges that arise from the advent of personal genomics services is to efficiently couple individual data with state of the art Pharmacogenomics (PGx) knowledge. Existing services are limited to either providing static views of PGx variants or applying a simplistic match between individual genotypes and existing PGx variants. Moreover, there is a considerable amount of haplotype variation associated with drug metabolism that is currently insufficiently addressed. Here, we present a web-based electronic Pharmacogenomics Assistant (ePGA; http://www.epga.gr/) that provides personalized genotype-to-phenotype translation, linked to state of the art clinical guidelines. ePGA's translation service matches individual genotype-profiles with PGx gene haplotypes and infers the corresponding diplotype and phenotype profiles, accompanied with summary statistics. Additional features include i) the ability to customize translation based on subsets of variants of clinical interest, and ii) to update the knowledge base with novel PGx findings. We demonstrate ePGA's functionality on genetic variation data from the 1000 Genomes Project.

Fig 1. The PGx haplotype table of gene UGT1A5 (as curated by PharmGKB).

Rows hold the variants, columns the PGx haplotypes, and each cell the major or minor allele used to define the corresponding haplotype. Shaded cells indicate the minor alleles.

Fig 2. Pharmacogenes in the Human Genome.

Dark grey bars refer to the relative number of PGx variants and grey bars to the relative number of haplotypes found in ePGA translation tables. A link between two pharmacogenes is added if these are involved in the metabolism of at least one common drug.

Fig 3. Number of haplotypes vs. number of variants for the 69 PGx genes.

Red dotted line denotes equal number of haplotypes and variants.

Fig 4. A workflow of user-ePGA interaction.

Fig 5. Translation process.

(A) Haplotype table for the UGT1A5 as downloaded from PharmGKB. (B) The modified UGT1A5 haplotype table by completing Step 1 (i,ii,iii sub-steps). (C) Numerical form of the UGT1A5 haplotype table. (D) UGT1A5 diplotypes (all combinations of haplotypes) with their numerically coded values. (E) An example of PGx variant annotation table with major and minor alleles as stated by PharmGKB. (F) A sample genotype profile. (G) The transformed, numerically coded, sample genotype profile described in Step 3. (H) The numerically coded sample diplotype to be matched.

Fig 6. Translation table for TPMT.

(A) Haplotype "rs1800460, rs2842934" is added named after its two variants. (B) Translation table for ASIC2 after the discovery of rs11869731 variant. (C) PGx report for individual HG00096, with a genotype of C/C in rs11869731, associated to an abnormal phenotype status.