RNA velocity estimation in R
Clone or download
pkharchenko Merge pull request #43 from Puriney/master
Enable velocyto.R support both hdf5r and h5
Latest commit 666e1db Nov 23, 2018



RNA velocity estimation in R

System requirements

velocyto.R can be installed on unix-flavored systems, and requires the following key elements:

  • C++11
  • Open MP support
  • boost libaries
  • igraph library
  • hdf5c++ library (as required by the h5 R package to support loom files)


The easiest way to install velocyto.R is using devtools::install_github() from R:


You need to have boost (e.g. sudo apt-get install libboost-dev) and openmp libraries installed. You can see detailed installation commands in the dockers/debian9/Dockerfile.


If you are having trouble installing the package on your system, you can build a docker instance that can be used on a wide range of systems and cloud environments. To install docker framework on your system see installation instruction. After installing the docker system, use the following commands to build a velocyto.R docker instance:

cd velocyto.R/dockers/debian9
docker build -t velocyto .
docker run --name velocyto -it velocyto


Chromaffin / SMART-seq2

The example shows how to annotate SMART-seq2 reads from bam file and estimate RNA velocity.

Dentate Gyrus / loom

The example shows how to load spliced/unspliced matrices from loom files prepared by velocyto.py CLI, use pagoda2 to cluster/embed cells, and then visualize RNA velocity on that embedding.

Mouse BM / dropEst

This example shows how to start analysis using dropEst count matrices, which can calculated from inDrop or 10x bam files using dropEst pipeline. It then uses pagoda2 to cluster/embed cells, and then visualize RNA velocity on that embedding.