ProjectSVR

ProjectSVR is a machine learning-based algorithm for mapping the query
cells onto well-constructed reference atlas.

Citation

ProjectSVR has been published in Briefings in Bioinformatics:

Jianing Gao, et al. (2025). "ProjectSVR: mapping single-cell RNA-seq data to reference atlases by supported vector regression." Briefings in Bioinformatics, Volume 26, Issue 6, bbaf586.
https://doi.org/10.1093/bib/bbaf586

Related resources

Quick start dataset

The data of quick
start
for ProjectSVR is available at https://zenodo.org/record/8147304 or
百度云盘(https://pan.baidu.com/s/13qSpcaldyQ9MUVCIaYSUIQ) 提取码: psvr

Reference atlas

The reference cell atlases involved in ProjectSVR paper are available at
https://zenodo.org/record/8350746 or
百度云盘(https://pan.baidu.com/s/1fNG5PcgqWiPZi3erkewA5w) 提取码: psvr

Query dataset

The query datasets involved in ProjectSVR paper are available at
https://zenodo.org/record/8350748 or
百度云盘(https://pan.baidu.com/s/1yGdhcwBIxodinRpppPHkQw) 提取码: psvr

Pre-built reference model

You can download pre-build reference models from
Zenodo or
百度云盘(https://pan.baidu.com/s/1yBWifQHimRNun1jgcYVEPg) 提取码: psvr

Tutorials

The ProjectSVR webpage with all the documentation and tutorials is
here.

We have various examples, including:

A generic quick start tutorial on a demo PBMC scRNA-seq dataset.

• Quick start: mapping PBMC dataset onto pre-build PBMC
  reference.

Tutorials on how to build projection models for reference atlas.

• DISCO blood
  atlas

• Maternal-fetal interface
  atlas

• Pan-cancer T cell
  lanscape

• Mouse testicular cell
  atlas

Tutorials on how to project the query datasets onto reference atlas via
pre-build models.

• Decidual immune microenvironment of RPL
  patients

• ICB responsiveness of
  BRCA

• Genetic perturbed germ cells
  (Zfp541-KO)

• Genetic perturbed germ cells
  (Ythdc2-KO)

• In vitro induced
  meiosis

A tutorial on how to train a model to predict pseudotime.

• Training the reference pseudotime
  model

A tutorial on how to train a multi-classifier for cell type auto
annotation.

• Training an ensemble SVM model for cell type
  prediction

Installation

Install the development version from GitHub use:

install.packages("devtools")
devtools::install_github("JarningGau/ProjectSVR")

ProjectSVR has been successfully installed and test on ubuntu, centOS
and wsl2.

Dependencies

• R >= 4.1

External packages

Install AUCell or UCell for signature score calculation.

## install UCell
# R = 4.3
BiocManager::install("UCell") # or
# R < 4.3
remotes::install_github("carmonalab/UCell", ref="v1.3")
## install AUCell
BiocManager::install("AUCell")

We provided a wrapper
RunCNMF
of python pacakge cnmf for feature
selection. If you want to use it, you should install cnmf through
reticulate.

install.packages("reticulate")
reticulate::install_miniconda()
## install sceasy for single cell data format transformation.
devtools::install_github("cellgeni/sceasy")
reticulate::py_install("anndata")
## install cnmf package via reticulate
reticulate::py_install("cnmf")

Benchmark results

Benchmark results of ProjectSVR and other reference mapping algorithms
were listed at https://github.com/JarningGau/ProjectSVR-benchmark/

Code of Conduct

Please note that the ProjectSVR project is released with a Contributor
Code of
Conduct.
By contributing to this project, you agree to abide by its terms.