a Julia library to normalize scRNAseq expression data and discover its low-dimensional latent space parameterization
SeqSpace provides both a Julia library, as well as a command line interface, to both normalize scRNAseq data and learn, if applicable, the underlying low-dimensional geometry. Our methodology is intended to be used on datasets that are well described by a small number of continuous degrees of freedom and thus represents an orthogonal viewpoint to that usually taken by traditional cell atlases. The inference from gene expression to continuous variables can be accomplished in either a supervised fashion, by mapping to a user-supplied database, or via an unsupervised approach that relies upon a novel machine learning architecture. SeqSpace is a standalone tool that we anticipate will be useful in analyzing scRNAseq data of developing systems.
The core algorithm and command line tools are self-contained and require no additional dependencies. The library is written in and thus requires Julia to be installed on your machine. Julia binaries for all operating systems can be found here.
Clone the repository
git clone https://github.com/nnoll/seqspace.git && cd seqspace
Build the package. This will create a separate Julia environment for SeqSpace
julia --project=. -e 'using Pkg; Pkg.build()'
Enter the REPL
Important please do not mix this method with that described above. Instead of creating a local SeqSpace specific environment, this method will install into the Julia base environment. We recommend, unless for a specific reason, to default to installing within a local environment. However, if needed, global installation can be achieved by running
julia -e 'using Pkg; Pkg.add(url="https://github.com/nnoll/seqspace.git")'
The SeqSpace package will now be available globally within the Julia REPL.