Imputation method for scRNA-seq based on low-rank approximation
Switch branches/tags
Nothing to show
Clone or download
Type Name Latest commit message Commit time
Failed to load latest commit information.
data first commit Aug 22, 2018 add info for alraSeurat2 function Nov 13, 2018
alra.R fix normalize function bug Nov 13, 2018
alraSeurat2.R alra with Seurat2 object Nov 13, 2018
alra_test.R added plotting to the test code Aug 23, 2018

Adaptively-thresholded Low Rank Approximation (ALRA)


ALRA is a method for imputation of missing values in single cell RNA-sequencing data, described in the preprint, "Zero-preserving imputation of scRNA-seq data using low-rank approximation" available here. Given a scRNA-seq expression matrix, ALRA first computes its rank-k approximation using randomized SVD. Next, each row (gene) is thresholded by the magnitude of the most negative value of that gene. Finally, the matrix is rescaled.

ALRA schematic

This repository contains codes for running ALRA in R. The only prerequisite for ALRA is installation of the randomized SVD package RSVD which can be installed as install.packages('rsvd').


Please be sure to pass ALRA a matrix where the cells are rows and genes are columns.

ALRA can be used as follows:

# Let A_norm be a normalized expression matrix where cells are rows and genes are columns.
# We use library and log normalization, but other approaches may also work well.
result.completed <- alra(A_norm)
A_norm_completed <- result.completed[[3]]

See alra_test.R for a complete example.

ALRA is integrated into Seurat v3.0 (currently at pre-release stage) as function RunALRA(). But if you use Seurat v2, we provide a simple function to perform ALRA on a Seurat v2 object in alraSeurat2.R.