Package index • Matisse

Creating a Matisse object

Start here. This function combines your Seurat object with splicing data into a single Matisse object (and computes PSI) in one call. Pass junction_counts for short-read (junction) mode, or transcript_counts for long-read (transcript) mode, plus splice events via events (data.frame or path to a SUPPA2 .ioe file).

show(<MatisseObject>) dim(<MatisseObject>) `[`(<MatisseObject>,<ANY>,<ANY>,<ANY>) `[[`(<MatisseObject>,<ANY>,<ANY>) `$`(<MatisseObject>): The MatisseObject S4 class
CreateMatisseObject(): Create a MatisseObject

Retrieve or update your data

Matisse-specific accessors. For raw count and annotation matrices, use the native Seurat API on the embedded Seurat object: e.g. SeuratObject::GetAssayData(GetSeurat(obj)[[“isoform”]], “counts”) for junction or transcript counts, or GetSeurat(obj)[[“psi”]][[]] for the per-event annotation table.

GetSeurat(): Get the embedded Seurat object
GetPSI(): Get the PSI matrix
SetPSI(): Set the PSI matrix
MatisseMeta() `MatisseMeta<-`(): Get or set cell-level metadata

Normalisation

Normalise count data before clustering. SCTransform applies variance stabilisation and scales for sequencing depth – in transcript mode it targets the isoform assay automatically; in junction mode it targets the gene-expression assay. NormalizeData, ScaleData, and FindVariableFeatures are log-normalisation alternatives.

SCTransform(<MatisseObject>): SCTransform normalisation for MatisseObjects
NormalizeData(<MatisseObject>): Normalise gene-expression counts for a MatisseObject
ScaleData(<MatisseObject>): Scale gene-expression data for a MatisseObject
FindVariableFeatures(<MatisseObject>): Identify highly variable features for a MatisseObject

Dimensionality reduction

Reduce the high-dimensional feature space to a compact representation before clustering and visualisation. RunPCA is the standard route after SCTransform; RunSVD (LSI) is used for ATAC-seq data in multiome experiments.

RunPCA(<MatisseObject>): Run PCA on a MatisseObject
RunUMAP(<MatisseObject>): Run UMAP on a MatisseObject
RunTSNE(<MatisseObject>): Run t-SNE on a MatisseObject

Clustering and differential expression

Group cells by transcriptional state and identify marker genes. All functions operate on the embedded Seurat object and return the updated MatisseObject — except FindMarkers, which returns a data frame.

FindNeighbors(<MatisseObject>): Compute a shared nearest-neighbour graph for a MatisseObject
FindClusters(<MatisseObject>): Cluster cells in a MatisseObject
FindMarkers(<MatisseObject>): Find differentially expressed markers for a MatisseObject

Metadata

Add or update per-cell metadata columns. New columns are accessible immediately via MatisseMeta() and the $ operator.

AddMetaData(<MatisseObject>): Add metadata columns to a MatisseObject

Signac / ATAC-seq methods

For multiome (paired ATAC + RNA) datasets. RunTFIDF normalises peak counts; FindTopFeatures selects variable peaks; RunSVD performs Latent Semantic Indexing (LSI) — the ATAC equivalent of PCA.

RunTFIDF(<MatisseObject>): Run TF-IDF normalisation for a MatisseObject
FindTopFeatures(<MatisseObject>): Find highly variable ATAC-seq features for a MatisseObject
RunSVD(<MatisseObject>): Run SVD (LSI) for a MatisseObject

Calculate splicing ratios (PSI)

Compute PSI (Percent Spliced In) – the fraction of transcripts in each cell that include a given exon. Values range from 0 (exon always skipped) to 1 (exon always included). CreateMatisseObject() calls CalculatePSI() automatically; you only call it directly to recompute with different parameters (e.g. min_coverage) or after constructing with defer_psi = TRUE.

CalculatePSI(): Calculate PSI matrix from junction or transcript counts
SummarizePSI(): Summarize PSI distribution across cells for each event

Quality control and filtering

Identify and remove cells or splicing events that don’t have enough data for reliable analysis. QC columns written automatically: nCount_isoform and nFeature_isoform at construction; nPercent_isoform after CalculatePSI().

FilterCells(): Filter cells by QC thresholds
FilterEvents(): Filter splice events by coverage or variance

Visualisation

Plot splicing patterns across your cells. Overlay any feature — PSI values, junction counts, or gene expression — on a UMAP, compare splicing between cell types, or inspect junction usage for a gene of interest. Pass the feature name via the feature argument.

PlotUMAP(): UMAP plot coloured by any feature
PlotViolin(): Violin plot of feature values split by cell group
PlotHeatmap(): Heatmap of PSI values (events x cells, DoHeatmap style)
PlotSashimi(): Sashimi-style coverage plot for a splice event

Utilities

Helper functions for combining datasets.

MergeMatisse(): Merge two MatisseObjects by cells