Pharmacogenomic Analysis of Drug Combination Experiments to Identify Biomarkers of Response or Resistance
Jermiah J. Joseph
Princess Margaret Cancer Centrejermiah.joseph@uhn.ca Source:
vignettes/PGx_2024.Rmd
PGx_2024.RmdWorkshop Overview
Instructor(s) name(s) and contact information
- Jermiah J. Joseph jermiah.joseph@uhn.ca
Workshop Description
This workshop will introduce users to the CoreGx and PharmacoGx R packages, which are useful tools for pharmacogenomic modelling to discover biomarkers of treatment response in cancer model systems. PharmacoGx specifically focuses on drug sensitivity experiments in cancer cell lines, which will be the major focus of this workshop. Additional infrastructure from our lab includes ToxicoGx for toxicogenomics in healthy human cell-lines, RadioGx for radiogenomics in cancer cell-lines and Xeva for pharmacogenomics in patient derived xenograph (PDX) murine models.
Participants will learn the fundamentals of using CoreGx and
PharmacoGx
to create a PharmacoSet—an integrative container for the
storage, analysis and visualization of pharmacogenomic experiments.
Particular focus will be placed on newly developed support for storing,
analyzing and visualizing drug combination sensitivity experiments and
correlating results therefrom with multi-omic molecular profiles to
discover biomarkers of drug sensitivity, resistance, synergy, or
antagonism.
Pre-requisites
- Basic knowledge of R syntax
- Knowledge of or interest in pharmacogenomics
- Familiarity with core Bioconductor classes such as the and SummarizedExperiment and MultiAssayExperiment
- Curiosity about or experience with data.table an asset!
Useful publications:
- Smirnov, P., Safikhani, Z., El-Hachem, N., Wang, D., She, A., Olsen, C., Freeman, M., Selby, H., Gendoo, D. M. A., Grossmann, P., Beck, A. H., Aerts, H. J. W. L., Lupien, M., Goldenberg, A., & Haibe-Kains, B. (2016). PharmacoGx: An R package for analysis of large pharmacogenomic datasets. Bioinformatics (Oxford, England), 32(8), 1244–1246. https://doi.org/10.1093/bioinformatics/btv723
- Tonekaboni, M., Ali, S., Soltan Ghoraie, L., Manem, V. S. K. & Haibe-Kains, B. Predictive approaches for drug combination discovery in cancer. Brief Bioinform 19, 263–276 (2018).
Introduction to CoreGx and PharmacoGx
This tutorial, titled Pharmacogenomic Analysis of Drug
Combination Experiments to Identify Biomarkers of Response or
Resistance, focuses on using the PharmacoGx
R package to correlate treatment response, measured as the viability of
cancer cell-lines after in vitro drug treatment, with their
respective multi-omic profiles. CoreGx
provides the core infrastructure for storing, analyzing, and visualizing
generic treatment response experiments. It provides methods and classes
which can be inherited from in downstream packages, such as ToxicoGx
and RadioGx.
We hope that the CoreSet object is generalized enough that
it can be reused by other developers for their specific treatment (or
stimuli) response use case.
CoreGx
Package Nomenclature
To facilitate modularization of the GxSuite of R packages, we have
shifted the nomenclature within a CoreSet—and therefore in
inheriting packages —to be more general.
To this end, have made the following changes:
- Previous reference to cell (cell-line) have become sample, allowing
the
CoreSetto be used for other model systems - Drug (radiation in
RadioGx) have become treatment, allowing theCoreSetto be treatment type (or stimuli) agnostic - Sensitivity will become response (sensitivity slot becomes treatmentResponse)
As a result of these changes, the names of some common accessors have been updated. The old accessors still remain functional to ensure backwards compatibility for at least two Bioconductor releases. A deprecation warning will be added to old accessors informing users of the corresponding new function, as per Bioconductor best practices.
PharmacoGx
PharmacoGx
stores drug screening data together with molecular profiling of
cell-lines in an object called a PharmacoSet, or
PSet for short.
Previously, the PharmacoSet class was entirely defined
by the PharmacoGx package. However, after building the
RadioGx and ToxicoGx packages, we realized
that the core data structures could be abstracted out and shared.
As such, PharmacoSets now inherit from the
CoreSet class defined in our package CoreGx,
which is used to share common datas structure and method across our
suite of package. The primary use case for PharmacoGx is to
:
- Provide a standardized and highly curated container for high-throughput screens in cancer-cell lines
- Enable discovery of biomarkers of treatment response or resistance
- Allow for comparison and validation across large published pharmacogneomic datasets
Overview of Data Structures
The GxSuite of packages make use of various Bioconductor classes for storing molecular profile data.
CoreSet and PharmacoSet
CoreSet are enclosed
in boxes. The first box indicates the type and name of each object. The
second box indicates the structure of an object or class. The third box
shows accessor methods from PharmacoGx for that specific
object. ‘=>’ represents return and specifies what is returned from
that item or method.Example Experiment
Cell Lines (samples)
The following cell lines are used in the study:
- Cell Line A (CLC-123): A breast cancer cell line.
- Cell Line B (CLC-456): A lung cancer cell line.
- Cell Line C (CLC-789): A colon cancer cell line.
- Cell Line D (CLC-101): A melanoma cell line.
- Cell Line E (CLC-202): A pancreatic cancer cell line.
Drugs (treatments)
The following drugs were tested on the cell lines:
- Drug X (Toxo-1): A novel chemotherapy drug that targets fast-dividing cancer cells.
- Drug Y (Inhi-2): A targeted therapy designed to block a protein that is commonly overactive in certain cancers.
- Drug Z (Syner-3): A combination therapy that enhances the effectiveness of Drug X when used together.