This repository contains scran.js, a Javascript library for single-cell RNA-seq (scRNA-seq) analysis in the browser. The various calculations are performed directly by the client, allowing us to take advantage of the ubiquity of the browser as a standalone analysis platform. Users can then directly analyze their data without needing to manage any dependencies or pay for access to a backend. scran.js is heavily inspired by the scran R package and contains most of its related methods. Indeed, much of the implementation in this repository is taken directly from scran and its related R packages.

Currently, the library and web app supports the key steps in a typical scRNA-seq analysis:

• Quality control (QC) to remove low-quality cells. This is done based on detection of outliers on QC metrics like the number of detected genes.
• Normalization and log-transformation, to remove biases and mitigate the mean-variance trend. We use scaling normalization with size factors defined from the library size for each cell.
• Feature selection to identify highly variable genes. This is based on residuals from a trend fitted to the means and variances of the log-normalized data for each gene.
• Principal components analysis (PCA) on the highly variable genes, to compress and denoise the data. We use an approximate method to quickly obtain the top few PCs.
• Clustering using multi-level community detection (a.k.a., "Louvain clustering"). This is performed on the top PCs.
• Dimensionality reduction with t-stochastic neighbor embedding (t-SNE), again using the top PCs.
• Marker detection using a variety of effect sizes such as Cohen's d and the area under the curve (AUC).

Coming soon:

• Clustering using k-means.
• Dimensionality reduction by uniform map and approximate projection (UMAP).
• Batch correction via the mutual nearest neighbors method.

The theory behind these methods is described in more detail in the Orchestrating Single Cell Analysis with Bioconductor book.

We use WebAssembly (Wasm) to enable efficient client-side execution of common steps in a scRNA-seq analysis. Code to perform each step is written in C++ and compiled to Wasm using the Emscripten toolchain. Some of the relevant C++ libraries are listed below:

• libscran provides C++ implementations of key functions in scran and its fellow packages scater and scuttle. This includes quality control, normalization, feature selection, PCA, clustering and dimensionality reduction.
• tatami provides an abstract interface to different matrix classes, focusing on row and column extraction.
• knncolle wraps a number of nearest neighbor detection methods in a consistent interface.
• CppIrlba contains a C++ port of the IRLBA algorithm for approximate PCA.
• CppKmeans contains C++ ports of the Hartigan-Wong and Lloyd algorithms for k-means clustering.
• qdtsne contains a refactored C++ implementation of the Barnes-Hut t-SNE dimensionality reduction algorithm.
• umappp contains a refactored C++ implementation of the UMAP dimensionality reduction algorithm.

For each step, we use Emscripten to compile the associated C++ functions into Wasm and generate Javascript-visible bindings. We can then load the Wasm binary into a web application and call the desired functions on user-supplied data.

This directory contains the files required to create the scran.js Wasm binary. We use CMake to manage the compilation process as well as the dependencies, namely the scran C++ library. Compilation of the Wasm binary is done using Emscripten:

emcmake cmake -S . -B build
(cd build && emmake make)

This will build the .js and .wasm file within the build/ subdirectory.