UINMF performs mosaic integration of single-cell multi-omic datasets using nonnegative matrix factorization

Abstract

Single-cell genomic technologies provide an unprecedented opportunity to define molecular cell types in a data-driven fashion, but present unique data integration challenges. Many analyses require “mosaic integration”, including both features shared across datasets and features exclusive to a single experiment. Previous computational integration approaches require that the input matrices share the same number of either genes or cells, and thus can use only shared features. To address this limitation, we derive a nonnegative matrix factorization algorithm for integrating single-cell datasets containing both shared and unshared features. The key advance is incorporating an additional metagene matrix that allows unshared features to inform the factorization. We demonstrate that incorporating unshared features significantly improves integration of single-cell RNA-seq, spatial transcriptomic, SNARE-seq, and cross-species datasets. We have incorporated the UINMF algorithm into the open-source LIGER R package (https://github.com/welch-lab/liger).

scMoMaT: Mosaic integration of single cell multi-omics data using matrix tri-factorization

“…To compare scMoMaT with baseline methods, we ran two recently published methods which can also work with this integration scenario: MultiMap 21 and UINMF 16 (Methods, visualization of latent embedding in Figs. S1a,b), and quantitatively measured the overall performance of all four methods with Graph connectivity (GC) score, NMI score, and ARI score (cite, Methods) using the label in the original data paper as ground truth.…”

“…It does not discuss how to utilize the information from all modalities for the batches that are measured with more than one modality. UINMF 16 uses a matrix bi-factorization framework to integrate data matrices with shared and unshared features. UINMF works with most mosaic integration cases, but it focuses on learning cell embedding and does not simultaneously learn a feature embedding along with the marker features of cell identities.…”

MOJITOO: a fast and universal method for integration of multimodal single-cell data

“…Both approaches explore efficient algorithms, but do not explicitly provide models associating molecular features to the ‘latent space’. MOFA ( Argelaguet et al , 2020 ), scAI ( Jin et al , 2020 ), totalVI ( Gayoso et al , 2021 ) and LIGER ( Kriebel and Welch, 2022 ) explore distinct methods for matrix factorization and estimation of shared latent spaces between modalities. Moreover, estimated matrices can be used for model interpretation, i.e.…”

One-Way Matching of Datasets with Low Rank Signals

“…To achieve such goals, it is necessary to (identify and) align cells in comparable states across related datasets. Yet another important application is the transfer and integration of complementary biological information across datasets: for example, if one dataset contains individual cells' spatial information within a tissue, matching it with a non-spatial single-cell dataset bears the potential of transferring spatial information to a different measurement modality (e.g., [43,23]).…”