FurtherMNN2018

Further MNN algorithm development

Overview

This repository contains code for further development of the mutual nearest neighbours (MNN) batch correction method, as implemented in the mnnCorrect and fastMNN functions in the batchelor package. It is based on the code at https://github.com/MarioniLab/MNN2017, which accompanies the paper Batch effects in single-cell RNA-sequencing data are corrected by matching mutual nearest neighbors by Haghverdi et al. (2018).

• Description of the theory behind fastMNN()
• Discussion of different normalization strategies
• Discussion of known failure points of fastMNN()
• Description of the clusterMNN() method

More details on practical usage can be found in the documentation for each function, as well as in the OSCA book - specifically in the Data Integration chapter and several workflows like the merged pancreas workflow.

Simulations

To run the simulations, enter the simulations/ directory and run:

• cluster_sim.R, which simulates a variety of scenarios involving orthogonal batch effects.
• nonorth_sim.R, which simulates some pathological non-orthogonal batch effects.

Real data

Three real data analyses are available - haematopoiesis, pancreas and droplet. Each subdirectory will usually contain:

• prepare.Rmd, to download, pre-process and normalize the data.
• correct.Rmd, to perform the batch correction and visualize the result with t-SNE plots.

Data file downloads are performed using the BiocFileCache package to save time and bandwidth. This will cache the files locally after the initial download, and reuse the cached versions when the script is re-run.