Variance-sensitive clustering of omics data

Feature-based variance-sensitive clustering of omics data. Optimizes cluster assignment by taking into account individual feature variance.

VSClust is available as interactive and user-friendly Shiny webservice and as R package in Bioconductor. For more instructions on both, see below.

VSClust was developed at the

Protein Research Group
Department of Biochemistry and Molecular Biology
University of Southern Denmark

Citation

When using VSClust, please cite our paper:
Veit Schwämmle, Ole N Jensen; VSClust: Feature-based variance-sensitive clustering of omics data, Bioinformatics, bty224, https://doi.org/10.1093/bioinformatics/bty224

Shiny app

Web service

You can use the implementation on our web server http://computproteomics.bmb.sdu.dk:
http://computproteomics.bmb.sdu.dk/Apps/VSClust

Be aware that the tool does allow only one user to run the background R calculations at a time. Therefore the app might become temporarily irresponsive. However, multiple sessions are separated and your data won't be shared between sessions or overwritten.

Local implementation

Docker

The easiest option is to use the docker image:

docker pull veitveit/vsclust
docker run -t -i -p 3838:3838 veitveit/vsclust

and access the server through http://localhost:3838

Bioconda

Install the package in the command-line

conda install -c bioconda vsclust
run_vsclust_app.sh

and access the shiny app through http://localhost:3838

Manual installation

Install the Bioconductor package vsclust In R:

if (!require("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install("vsclust")

You can run the shiny app from the server.R or ui.R files in the "inst/shiny" folder using Rstudio or run the app on a shiny-server

Be aware that you need to have all files, the R libraries described in Installation and the modified e1071 library installed.

Build and use Docker image

A Dockerfile has been created on top of the rocker/shiny docker image. Copy this repository to a folder and carry out the following command to build the image (takes a while)

docker build -t veitveit/vsclust .

You can also just directly download and run the image by

docker run -t -i -p 3838:3838 veitveit/vsclust

and access the server through http://localhost:3838

Command line and R package

All operations but the gene set enrichment can be performed via command line running the R script runVSClust.R or using the functions of the Bioconductor package vsclust

Installation

In R:

if (!require("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install("vsclust")

As conda package

TBD

Usage

Please take a look at the vignettes and/or in the help packages of the vsclust functions

Contact

For software issues and general questions, please submit an issue.

License

GPL-2 or higher