Skip to contents

Create a MatisseObject

Source: R/constructors.R
CreateMatisseObject.Rd

The single constructor for MatisseObject. Combines a Seurat object with isoform-resolved splicing data and also computes PSI in one step (unless defer_psi = TRUE). The operating mode is detected automatically from the inputs you supply:

Usage

CreateMatisseObject(
  seurat,
  junction_counts = NULL,
  transcript_counts = NULL,
  events = NULL,
  min_coverage = 5L,
  defer_psi = FALSE,
  verbose = TRUE
)

Arguments

seurat

A Seurat object. Required.

junction_counts

A sparse matrix (dgCMatrix, cells x junctions) of raw per-junction read counts. Row names must match colnames(seurat). Junction IDs (column names) ideally encode genomic coordinates (chr-start-end-strand, chr:start-end:strand, or chr_start_end_strand); Matisse parses these for sashimi plots. Triggers junction mode. Default: NULL.

transcript_counts

A matrix or sparse matrix (transcripts x cells) of raw transcript-level counts. Stored as Assay5("isoform") in the Seurat object. Column names must overlap with colnames(seurat). Triggers transcript mode. Default: NULL.

events

Splice-event annotation. Either:

  • a character vector of paths to SUPPA2 .ioe file(s) (parsed internally), or

  • a pre-built data.frame with columns event_id, gene_id, chr, strand, event_type, inclusion_features, exclusion_features. The *_features columns hold the IDs that support each role — junction IDs in junction mode, transcript IDs in transcript mode.

Required.

min_coverage

Integer. Minimum total reads per cell per event for the PSI calculation step. Default: 5L. Forwarded to CalculatePSI when defer_psi = FALSE.

defer_psi

Logical. Skip the PSI calculation step at construction. The returned object will have raw counts only (no "psi" assay). Call CalculatePSI later. Default: FALSE.

verbose

Logical. Print construction progress. Default: TRUE.

Value

A MatisseObject.

Details

  • Junction mode (short-read): pass junction_counts plus splice events via events. Junction counts are stored as Assay5("isoform") inside the Seurat object; PSI values land in Assay5("psi").

  • Transcript mode (long-read): pass transcript_counts plus splice events via events (typically a SUPPA2 IOE file). Transcript counts are stored as Assay5("isoform"); PSI values land in Assay5("psi").

Matisse intentionally does not ship a heuristic event caller — bring your own splice-event annotation (SUPPA2 IOE, rMATS table, or equivalent).

Examples

if (FALSE) { # \dontrun{
library(Seurat)
counts <- matrix(rpois(200, 5), nrow = 20,
                 dimnames = list(paste0("Gene", 1:20),
                                 paste0("Cell", 1:10)))
seu <- CreateSeuratObject(counts)

# Junction mode (short-read): events as a data.frame
jxn <- make_junction_counts()
obj <- CreateMatisseObject(seurat = seu, junction_counts = jxn,
                           events = my_event_df)

# Transcript mode (long-read): events as SUPPA2 IOE file paths
tx  <- make_transcript_counts()
obj <- CreateMatisseObject(seurat = seu, transcript_counts = tx,
                           events = "path/to/events.ioe")
} # }