scPharm

---

Overview

---

R package for pharmacological cell subpopulations identification, drug prioritization and potetial drug combinations identification from single-cell RNA-seq of tumour tissue. scPharm is a computational framework tailored for scRNA-seq data, integrating pharmacogenomics profiles to uncover therapeutic heterogeneity within tumours at single-cell resolution. The tool not only prioritizes tailored drugs but also provides insight into combination therapy regimen and drug toxicity in cancers.

Installation

---

Within R, set first:

install.packages("remotes")

To install scPharm then just type:

remotes::install_github("WangHYLab/scPharm", dependencies = T)

Data inputs format

---

scPharm use as input single cell data in the form of specific S4 objects.

Usage

---

Function examples:

Identify pharmacological cell subpopulations

scPharmIdentify.result <- scPharmIdentify(object = example_data, type = "tissue", cancer = "BRCA")

parameter of scPharmIdentify function

object: a seurat object of patient or cell line cells(example data can be download from https://github.com/WangHYLab/scPharm/blob/main/example_data.rds).
type: the source of cell, cell line or tumor tissue. Can be set to 'cellline' or 'tissue'.
cancer: the TCGA cancer type of cell. A character or vector.(eg: BRCA or c('LUAD', 'LUSC')). cancer='pan' means calculating in the context of pan-cancer.
drug: the drug name for identifying pharmacological cell subpopulations. If not specified all drugs from GDSC2 project will be taken.
nmcs: number of components to compute and store for MCA. Default:50
nfeatures: number of genes used to make cell ID.
cores: number of CPU cores to use. This parameter can only be set to 1 on windows platform. Default:1
features: character vector of feature names to run MCA. If not specified all features will be taken.
slot: slot of seurat object used to run MCA.
assay: assay of seurat object used to run MCA.
threshold.s: the threshold to label sensitive cells. Default: -1.751302
threshold.r: the threshold to label resistant cells. Default: 1.518551
bulkdata: file for all cancer cell lines, automatically loaded in scPharm.
gdscdata: pharmacological file for all cancer cell lines, automatically loaded in scPharm.

value

A seurat object added pharmacological label to cells about different drugs in GDSC2 project. Label for drugs in seurat object meta data named with 'scPharm_label_DURGID_DRUGNAME'.

Compute Dr(drug prioritization)

scPharm.Dr <- scPharmDr(scPharmIdentify.result)

parameter of scPharmDr function

object: a seurat object of patient or cell line cells after running scPharmIdentify function.

value

A data frame consisted of six columns. Column 1 and 2 are id and name of drugs in GDSC2 project. Column 3 and 4 are the size of sensitive and resistant subpopulations to specific drug. Column 5 and 6 are score(Dr) and prioritization to different drugs for specific patient.

Identify potential drug combinations

scPharm.combo <- scPharmCombo(scPahrmIdentify.result, scPharmDr.result)

parameter of scPharmCombo function

object: a seurat object after running scPharmIdentify function
score: output of scPharmDr function.
drug: name of drug as drug 1. Default:NULL
topN: number of drugs at top of Dr list as drug 1. Default:1 drug_info: file for all drugs, automatically loaded in scPharm.

value

A data frame consisted of five columns. Column 1 is name of drug 1. Column 2 and 3 are id and name of drug 2. Column 4 and 5 are the value of combiantion effects and the type of combination.

Predict drug side effects

scPharm.Dse <- scPharmDse(scPharmIdentify.result)

parameter of scPharmDse function

object: a seurat object of patient cells after running scPharmIdentify function.

value

A data frame consisted of three columns. Column 1 and 2 are id and name of drugs in GDSC2 project. Column 3 is the value of side effects to different drugs.

Generate the null distribution and calculate the thresholds

scPharmGenNullDist.result <- scPharmGenNullDist(object = example_data, cancer = "pan")

parameter of scPharmGenNullDist function

object: a seurat object of cells from healthy tissue.
cancer: the TCGA cancer type of cell. A character or vector.(eg: BRCA or c('LUAD', 'LUSC')). cancer='pan' means calculating in the context of pan-cancer. nmcs: number of components to compute and store for MCA. Default:50
nfeatures: number of genes used to make cell ID.
cores: number of CPU cores to use. This parameter can only be set to 1 on windows platform. Default:1
features: character vector of feature names to run MCA. If not specified all features will be taken.
slot: slot of seurat object used to run MCA.
assay: assay of seurat object used to run MCA.
bulkdata: file for all cancer cell lines, automatically loaded in scPharm.
gdscdata: pharmacological file for all cancer cell lines, automatically loaded in scPharm.

value

A list contains the null distribution(NullDist), thresholds for labeling sensitive and resistant cells(threshold_s and threshold_r).

References

---

Tian P, Zheng J, Xu Y, et al. scPharm: identifying pharmacological subpopulations of single cells for precision medicine in cancers. Published online December 12, 2023. doi:10.1101/2023.12.11.571182

---