Pharmacogx 3.0: Data Structures And Methods For Discovering Molecular Biomarkers Of Response To Drug Combination Therapy In Precision Oncology**

Pharmacogx 3.0: Data Structures And Methods For Discovering Molecular Biomarkers Of Response To Drug Combination Therapy In Precision Oncology**

Author(s): Christopher Bernard Eeles, Petr Smirnov, Benjamin Haibe-Kains

Affiliation(s): Princess Margaret Cancer Center

Initially published to address gaps in the reproducibility and comparability of large high-throughput pharmacogenomic studies, PharmacoGx has evolved into a rich set of tools for storing, annotating, curating, and analyzing the relationship between molecular phenotypes and drug response in preclinical cancer models. Public, version-controlled release of R PharmacoSet objects for studies such as GDSC and CCLE has enabled simple, reproducible access to high-quality pharmacogenomic data via PharmacoGx, allowing researchers to focus on interesting analyses which generate novel insights. To date, PharmacoGx has focused primarily on modelling responses to single agent chemo- and targeted therapies in cancer cell-lines. However, interest in predicting synergistic drug combinations for translation into clinical trials is increasing and promises to overcome limitations of monotherapy such as incomplete response or acquired resistance. To address this need, we have developed a new S4 class, the TreatmentResponseExperiment, which provides efficient storage of dose-response data for both mono and combination therapies. Integration of this new class into the PharmacoSet will allow storage and analysis of complex drug combination experiments and facilitate discovery of biomarkers for drug synergy or antagonism across a range of different omics technologies in cancer cell-lines and beyond. This workshop aims to be a comprehensive introduction to PharmacoGx, and will be suitable for both new and experienced package users. The presentation will begin with an overview of the PharmacoGx package, including review of relevant Bioconductor and PharmacoGx data structures. Particular focus will be placed on creating a TreatmentResponseExperiment (TRE) using published drug combination data, highlighting how various included helper methods can assist users in mapping from raw experiment data to the slots of a TRE. After TRE creation, we will demonstrate how the PharmacoSet (PSet) constructor can be used to create and annotate a dataset for downstream analysis. Users will have the opportunity to interact with us and ask questions during an interactive session to ensure they understand the process for creating a PSet, with the goal of teaching users the skills to create one from their own research data. This section will also highlight the anatomy of a PSet and get users familiar with the various accessor and subset methods they will need to effectively leverage a PSet in subsequent computational analyses. The second half of the workshop will begin with an introduction to various metrics of drug response—such as IC50, EC50, and AAC—and drug synergy—such as best average response, BLISS, and ZIP—as well as how they can be computed using PharmacoGx. We will introduce best practices for modelling the relationship between dose-response and molecular phenotypes and how methods implemented in PharmacoGx can be used to identify molecular features associated with response or resistance to a given (combination) therapy. A second interactive section will provide users the time to compute metrics of drug response, resistance, synergy and antagonism using our sample drug combination data as well as to apply the modelling methods in PharmacoGx to discover associated univariate biomarkers. Once users have completed this analysis, we will discuss considerations for interpreting discovered

Workshop details

Source code


1. Go to Orchestra.
2. Log in.
3. Search for the workshop of interest.
4. Click "Launch" (may take a minute or two).
5. Follow instructions.