It is assumed that human genetics can help identify new drug targets. However, the best way to prioritize genes as therapeutic targets remains controversial. Authors [see attached article & editorial] describe a framework to prioritize potential targets by integrating genome-wide association studies (GWAS) data with genomic architecture, development of diseases, and network connectivity. Although their genetics-led drug-target prioritization approach is focused on immune cell-mediated traits, this framework should also be applicable to non-immunologically-mediated diseases.
Authors state that there are two general approaches to prioritize genes from human genetic studies as therapeutic targets. The first is a gene-centric approach. One model takes advantage of trait-associated alleles (remember: an allele is one copy of the gene; the other allele represents the second copy of the gene on the other chromosome) to estimate dose–response curves; in this model, the trait-associated alleles could arise from common-variant association studies, rare-variant association studies, or studies of rare Mendelian phenotypes. For common diseases — examples include PCSK9 (proprotein convertase subtilisin/kexin type-9 in coronary artery disease) and TYK2 (tyrosine kinase-2 in immunologically mediated diseases), and for rare diseases, examples include CFTR [CF trans-membrane conductance regulator (ABCC7) in cystic fibrosis] and SMN1 (survival of motor neuron 1, telomeric in spinal muscular atrophy).
Another gene-centric model nominates individual genes that arise from GWAS, by using genomic features, and other annotations — such as phenotype hierarchy. A benefit of this ‘genomic-features’ model is that many GWAS signals arise from non-coding regions; therefore, prioritizing individual genes from an implicated region is difficult. An additional benefit of the genomic-features model is that it does not require more than a single trait-associated allele (which is usually the case for GWAS).
A second approach is to build networks, or pathways, based on connectivity among ‘seed genes’ implicated by human genetics and then to expand the network to include non-seed genes by using orthogonal data such as protein–protein interactions. Seed genes can originate from either the allelic-series model or the genomic-features model. An advantage of this pathway-centric approach is that many potential targets do not contain naturally occurring variants that disrupt gene function — yet they are still associated with a relevant trait.
Authors [see attached article] developed the priority index pipeline, taking (as inputs) GWAS variants for specific immune traits. These variants are predominantly regulatory, commonly act at a distance, and are often context-specific (pertaining to a distinct case). Authors used genomic predictors to identify and score genes likely to be responsible for GWAS signals (denoted as ‘seed genes’), based on: [a] genomic proximity to a disease-associated single-nucleotide variant (SNV), accounting for linkage disequilibrium (non-random association of alleles at different loci in a given population, due to inheritance) and genomic organization; [b] physical interaction, as evidenced by chromatin conformation in immune cells; and [c] modulation of gene expression, evidenced by expression quantitative trait loci (eQTL) in immune cells.
Authors demonstrate how their genetics-led drug-target prioritization approach (the priority index) successfully identifies current therapeutics, predicts activity in high-throughput cellular screens, enables prioritization of underexplored targets, and allows for determination of target-level trait relationships. The priority index is an open-access, scalable system accelerating early-stage drug target selection for immune-mediated disease. It should also be applicable for other forms of human complex diseases. 😊
Nat Genet July 2019; 51: 1082-1091 & pp 1073-1075 (News’N’Views editorial)