snifter

A collection of tools for doing various analyses of single-cell RNA-seq gene expression data, with a focus on quality control and visualization.

Implements the density-preserving modification to t-SNE and UMAP described by Narayan et al. (2020) <doi:10.1101/2020.05.12.077776>. The non-linear dimensionality reduction techniques t-SNE and UMAP enable users to summarise complex high-dimensional sequencing data such as single cell RNAseq using lower dimensional representations. These lower dimensional representations enable the visualisation of discrete transcriptional states, as well as continuous trajectory (for example, in early development). However, these methods focus on the local neighbourhood structure of the data. In some cases, this results in misleading visualisations, where the density of cells in the low-dimensional embedding does not represent the transcriptional heterogeneity of data in the original high-dimensional space. den-SNE and densMAP aim to enable more accurate visual interpretation of high-dimensional datasets by producing lower-dimensional embeddings that accurately represent the heterogeneity of the original high-dimensional space, enabling the identification of homogeneous and heterogeneous cell states. This accuracy is accomplished by including in the optimisation process a term which considers the local density of points in the original high-dimensional space. This can help to create visualisations that are more representative of heterogeneity in the original high-dimensional space.

Builds hexbin plots for variables and dimension reduction stored in single cell omics data such as SingleCellExperiment. The ideas used in this package are based on the excellent work of Dan Carr, Nicholas Lewin-Koh, Martin Maechler and Thomas Lumley.

omicsGMF is a Bioconductor package that uses the sgdGMF-framework of the \code{sgdGMF} package for highly performant and fast matrix factorization that can be used for dimensionality reduction, visualization and imputation of omics data. It considers data from the general exponential family as input, and therefore suits the use of both RNA-seq (Poisson or Negative Binomial data) and proteomics data (Gaussian data). It does not require prior transformation of counts to the log-scale, because it rather optimizes the deviances from the data family specified. Also, it allows to correct for known sample-level and feature-level covariates, therefore enabling visualization and dimensionality reduction upon batch correction. Last but not least, it deals with missing values, and allows to impute these after matrix factorization, useful for proteomics data. This Bioconductor package allows input of SummarizedExperiment, SingleCellExperiment, and QFeature classes.

Correspondence analysis (CA) is a matrix factorization method, and is similar to principal components analysis (PCA). Whereas PCA is designed for application to continuous, approximately normally distributed data, CA is appropriate for non-negative, count-based data that are in the same additive scale. The corral package implements CA for dimensionality reduction of a single matrix of single-cell data, as well as a multi-table adaptation of CA that leverages data-optimized scaling to align data generated from different sequencing platforms by projecting into a shared latent space. corral utilizes sparse matrices and a fast implementation of SVD, and can be called directly on Bioconductor objects (e.g., SingleCellExperiment) for easy pipeline integration. The package also includes additional options, including variations of CA to address overdispersion in count data (e.g., Freeman-Tukey chi-squared residual), as well as the option to apply CA-style processing to continuous data (e.g., proteomic TOF intensities) with the Hellinger distance adaptation of CA.

slalom is a scalable modelling framework for single-cell RNA-seq data that uses gene set annotations to dissect single-cell transcriptome heterogeneity, thereby allowing to identify biological drivers of cell-to-cell variability and model confounding factors. The method uses Bayesian factor analysis with a latent variable model to identify active pathways (selected by the user, e.g. KEGG pathways) that explain variation in a single-cell RNA-seq dataset. This an R/C++ implementation of the f-scLVM Python package. See the publication describing the method at https://doi.org/10.1186/s13059-017-1334-8.