PharmacoGx Embeddings

Use drug and gene embeddings to predict drug-gene interactions, with a focus on Africa.

Project overview

• Project progress is tracked via GitHub Projects.
• Check the PharmacoGx-ARSA repository for more details on the processing of 'African variants'.
• Find the app code in the PharmacoGX-App repository.
• Read our comment in Nature about the launch of this project.
• To learn more about the GRADIENT projects, see this article in Drug Discovery Today.

Knowledge graph

The knowledge graph leverages mainly PharmGKB and the Bioteque:

• PharmGKB is used as a source of pharmacogenomics knowledge. We use the different concepts summarized in PharmGKB as the backbones of the knowledge graph. These include, for example, evidence levels of the variant-drug relationships, or the biogeographical groups in which a variant has been identified.
• The Bioteque is a knowledge graph itself integrating relations between 12 biological entities, and describing them as ML-amenable descriptors. We use it to complement the knowledge graph with additional information.

Ontology

To capture the main pharmacogenomic relationships and concepts in our KG, we have adapted the structure of PharmGKB and the light-weight pharmacogenomics ontology (PGxO). The below graph illustrates the main concepts captured in the KG, as well as the edges that relate the different concepts. The central piece of the KG is the pharmacogenomic relationship, which, ideally, is able to describe the phenotypic effect a variant has on a specific drug. This interactions can range from well-proven (evidence levels 0 to 3), annotated relations to automated literature searches. In Yellow, we depict the Base Tables, which build upon PharmGKB's database. In Green, we depict relationship tables we have built based on the Base tables. The PGX Relationship table draws from all of them to integrate the different concepts.

graph LR
    A(Variant):::YellowClass --> C(Genomic Variation):::GreenClass
    B(Haplotype):::YellowClass --> C(Genomic Variation)
    C --> D(Genetic Factor):::GreenClass
    E(Gene):::YellowClass --> D
    F(Chemicals):::YellowClass --> G(Pharmacogenomic Relationship):::OrangeClass
    D --> G
    H(Evidence):::YellowClass --> G
    I(Biogeographical Group):::YellowClass --> G
    G --> J(Phenotype):::GreenClass
    J --> K(Pharmacodynamic phenotype):::YellowClass
    J --> L(Pharmacokinetic phenotype):::YellowClass
    J --> M(Disease):::YellowClass

    classDef OrangeClass fill:#faa08c,stroke:#50285a,stroke-width:2px
    classDef GreenClass fill:#bee6b4,stroke:#50285a,stroke-width:2px
    classDef YellowClass fill:#fad782,stroke:#50285a,stroke-width:2px

Curation of PharmGKB

Since PharmGKB is the major source of information for the KG, below we detail a few of the assumptions made during data curation and processing, to build the tables that can be found in the corresponding folder. This is simply an overview of the data curation process, please refer to the code for more details.

• Chemicals: each chemical, belonging to a specific type (Drug, Metabolite, Small Molecule, Ion...) is associated to a chemical ID (cid, PharmGKB Accession ID), a name, and a SMILES. If the SMILEs is not available, we discard the chemical for the current KG. The dosing guideline status (Yes: 1, No: 0) is also collected.
• Genes: each gene is associated to a gene ID (gid, PharmGKB Accession ID). For each gene, we define whether it is a VIP, has variants annotated or dosing guidelines (Yes: 1, No: 0)
• Variants: the variants table relates each variant, identified by its name and a variant ID, to the gene(s) where it belongs (gid).
• Haplotypes: haplotypes are identified in the relationships table from PharmGKB. We collect their ID, name, and associated gene(s).
• Diseases: diseases are listed from the phenotype files, indexed by ID (did, PharmGKB Accession ID) and name.
• PK phenotype: as defined by PharmGKB, pharmacokinetics phenotype can be Dosing, Metabolism/PK or Other. Relations without described phenotype are Nan
• PD phenotype: as defined by PharmGKB, pharmacodynamics phenotype can be Efficacy, Toxicity, PD or Other. Relations without described phenotype are Nan
• Evidence: the evidence levels indicate how reliable an association is. 0A: Actionable PGx (from drug labels), 0B: Informative PGx (from drug labels), 0C: Testing recommended (from drug labels), 0D: Testing required (from drug labels), 0E: VIP Gene, 0F: Pathway, 0G: Dosing Guideline, 1A, 1B, 2A, 2B, 3, 4: clinical annotations, , 5A: variant_drug annotations, 5B: variant_phenotype annotations, 5C: variant_fa annotations, 6A: automated annotations, 6B: DataAnnotations, 6C: Literature, 6D: Multilink Annotations
• Biogeographical group: described in PharmGKB

These tables are merged into the PGX relationship table. At this moment, the PGX mainly encompasses the following information: annotation ID (unique identifier of the relationship described), genomic variation (variant or haplotype) and its ID, gene and its ID, chemical and its ID, pk and pd phenotype, evidence level, association (significance of the association Yes: 1 / No: -1 / Inconclusive: 0) and biogeograhical group.

Use of Large Language Models (LLMs)

LLMs are used to re-rank and offer an explanation of the predictions. We mainly used gpt-4-turbo.

Repository structure

• assets: graphical assets for the project
• data: contains all the .csv and .tsv files, including those directly downloaded from open repositories
• scripts: contains .py scripts analyse the data
• notebooks: testing notebooks and examples
• src: source code for the building of the KG

Please note that several folders will be generated when you run the pipeline below, including embeddings, models and results folders. Also, please note that the data folder will be modified.

Experiment pipelines

• Input: PharmGKB drug-gene pairs.
• Output: All pair-wise combinations on PharmGKB genes and drugs.

1. Run scripts in scripts/0_process_pharmgkb:
2. Run scripts in scripts/1_preparation:
3. Run scripts in scripts/2_pairs to
4. Run scripts in scripts/3_llm_assessment to rerank drug-gene pair predictions based on large language models.
5. Run scripts on scripts/4_post_analyses to produce analyses on the results, including statistics and plots.

License

The code in this repository is licensed under a GNU General Public License v3.0. The data comes from public repositories and is limited to the licenses stated by the original data producers.

About Us

The Ersilia Open Source Initiative is a Non Profit Organization (1192266) with the mission is to equip labs, universities and clinics in LMIC with AI/ML tools for infectious disease research.

Help us achieve our mission or volunteer with us!