Run the full tinydenseR pipeline — RunTDR • tinydenseR

S3 generic that executes the complete tinydenseR analysis pipeline: landmark selection, graph construction, optional cell typing, and mapping.

Usage

RunTDR(x, ...)

# Default S3 method
RunTDR(
  x,
  .celltype.vec = NULL,
  .celltype.vec.overwrite = FALSE,
  .verbose = TRUE,
  .seed = 123,
  ...
)

# S3 method for class 'Seurat'
RunTDR(
  x,
  .sample.var,
  .assay = "RNA",
  .layer = "counts",
  .assay.type = "RNA",
  .harmony.var = NULL,
  .markers = NULL,
  .celltype.vec = NULL,
  .min.cells.per.sample = 10,
  .verbose = TRUE,
  .seed = 123,
  .prop.landmarks = 0.1,
  .n.threads = if (is.hpc()) {
     max(RhpcBLASctl::blas_get_num_procs(),
    RhpcBLASctl::omp_get_num_procs(), RhpcBLASctl::omp_get_max_threads(), na.rm = TRUE)

    } else {
     parallel::detectCores(logical = TRUE)
 },
  ...
)

# S3 method for class 'SingleCellExperiment'
RunTDR(
  x,
  .sample.var,
  .assay = "counts",
  .assay.type = "RNA",
  .bpcells.dir = NULL,
  .harmony.var = NULL,
  .markers = NULL,
  .celltype.vec = NULL,
  .min.cells.per.sample = 10,
  .verbose = TRUE,
  .seed = 123,
  .prop.landmarks = 0.1,
  .n.threads = if (is.hpc()) {
     max(RhpcBLASctl::blas_get_num_procs(),
    RhpcBLASctl::omp_get_num_procs(), RhpcBLASctl::omp_get_max_threads(), na.rm = TRUE)

    } else {
     parallel::detectCores(logical = TRUE)
 },
  ...
)

# S3 method for class 'HDF5AnnData'
RunTDR(
  x,
  .sample.var,
  .assay.type = "RNA",
  .h5ad.group = NULL,
  .bpcells.dir = NULL,
  .harmony.var = NULL,
  .markers = NULL,
  .celltype.vec = NULL,
  .min.cells.per.sample = 10,
  .verbose = TRUE,
  .seed = 123,
  .prop.landmarks = 0.1,
  .n.threads = if (is.hpc()) {
     max(RhpcBLASctl::blas_get_num_procs(),
    RhpcBLASctl::omp_get_num_procs(), RhpcBLASctl::omp_get_max_threads(), na.rm = TRUE)

    } else {
     parallel::detectCores(logical = TRUE)
 },
  ...
)

# S3 method for class 'character'
RunTDR(
  x,
  .sample.var,
  .assay.type = "RNA",
  .h5ad.group = NULL,
  .bpcells.dir = NULL,
  .harmony.var = NULL,
  .markers = NULL,
  .celltype.vec = NULL,
  .min.cells.per.sample = 10,
  .verbose = TRUE,
  .seed = 123,
  .prop.landmarks = 0.1,
  .n.threads = if (is.hpc()) {
     max(RhpcBLASctl::blas_get_num_procs(),
    RhpcBLASctl::omp_get_num_procs(), RhpcBLASctl::omp_get_max_threads(), na.rm = TRUE)

    } else {
     parallel::detectCores(logical = TRUE)
 },
  ...
)

# S3 method for class 'cytoset'
RunTDR(
  x,
  .sample.var,
  .assay.type = "cyto",
  .harmony.var = NULL,
  .markers = NULL,
  .celltype.vec = NULL,
  .min.cells.per.sample = 10,
  .verbose = TRUE,
  .seed = 123,
  .prop.landmarks = 0.1,
  .n.threads = if (is.hpc()) {
     max(RhpcBLASctl::blas_get_num_procs(),
    RhpcBLASctl::omp_get_num_procs(), RhpcBLASctl::omp_get_max_threads(), na.rm = TRUE)

    } else {
     parallel::detectCores(logical = TRUE)
 },
  ...
)

# S3 method for class 'flowSet'
RunTDR(
  x,
  .sample.var,
  .assay.type = "cyto",
  .harmony.var = NULL,
  .markers = NULL,
  .celltype.vec = NULL,
  .min.cells.per.sample = 10,
  .verbose = TRUE,
  .seed = 123,
  .prop.landmarks = 0.1,
  .n.threads = if (is.hpc()) {
     max(RhpcBLASctl::blas_get_num_procs(),
    RhpcBLASctl::omp_get_num_procs(), RhpcBLASctl::omp_get_max_threads(), na.rm = TRUE)

    } else {
     parallel::detectCores(logical = TRUE)
 },
  ...
)

# S3 method for class 'dgCMatrix'
RunTDR(x, .cell.meta, ...)

# S3 method for class 'DelayedMatrix'
RunTDR(x, .cell.meta, .bpcells.dir = NULL, ...)

# S3 method for class 'IterableMatrix'
RunTDR(x, .cell.meta, ...)

Arguments

x: A DelayedMatrix (features × cells for RNA).
...: Additional arguments passed to .run_tdr_matrix (e.g. .sample.var, .assay.type, .harmony.var, .markers, .celltype.vec, .min.cells.per.sample, .verbose, .seed, .prop.landmarks, .n.threads).
.celltype.vec: Character(1). Column name in .cell.meta containing per-cell type labels, or NULL.
.celltype.vec.overwrite: Logical. If TRUE, replace an existing .celltype.vec stored in the TDRObj config.
.verbose: Logical. Print progress messages.
.seed: Integer. Random seed.
.sample.var: Character(1). Column name in .cell.meta identifying sample membership.
.assay: Character(1). Name of the assay in assayNames(x).
.layer: Character(1). Layer within the assay (e.g. "counts").
.assay.type: Character. "RNA" or "cyto".
.harmony.var: Character vector of batch variable column names in sample-level metadata, or NULL.
.markers: Character vector of marker names (required for cyto).
.min.cells.per.sample: Integer. Minimum cells for a sample to be included.
.prop.landmarks: Numeric in (0, 1]. Proportion of cells as landmarks.
.n.threads: Integer. Number of threads.
.bpcells.dir: Character(1) or NULL. Directory path for the BPCells on-disk matrix. If NULL (default), uses a temporary directory (tempdir()) that is cleaned up on session end. If a path is given and already contains a valid BPCells matrix, the conversion step is skipped (cache hit).
.h5ad.group: Character(1) or NULL. HDF5 group path containing the count matrix (e.g. "/layers/counts" or "/X"). If NULL (default), auto-detects by probing /layers/counts first, then falling back to /X.
.cell.meta: A data.frame of per-cell metadata. Must have one row per cell with rownames matching cell IDs in x (colnames for RNA, rownames for cyto).

Value

Depends on the method; see individual method documentation.

The updated TDRObj.

The Seurat object x with the TDRObj stored in Misc(x, slot = "tdr.obj").

The SingleCellExperiment x with the TDRObj stored in S4Vectors::metadata(x)$tdr.obj.

A TDRObj.

The updated TDRObj.

A TDRObj.

Details

RunTDR is a convenience wrapper around the explicit step-by-step workflow (get.meta → get.cells → setup.tdr.obj → get.landmarks → get.graph → get.map → get.embedding). For equivalent inputs, seed, and arguments the two paths produce numerically identical results.

Methods (by class)

RunTDR(default): Run the pipeline on an existing TDRObj

Executes get.landmarks → get.graph → celltyping → get.map → get.embedding on a pre-built TDRObj.
RunTDR(Seurat): Run the pipeline directly on a Seurat object

Builds a TDRObj from a Seurat object and executes the full pipeline. The finished TDRObj is stored in SeuratObject::Misc(x, slot = "tdr.obj").

All categorical cell-level columns in x@meta.data are automatically imported as named celltyping solutions via import_cell_annotations. The .celltype.vec column (if specified) is set as the active annotation.
RunTDR(SingleCellExperiment): Run the pipeline directly on a SingleCellExperiment

Builds a TDRObj from a SingleCellExperiment and executes the full pipeline. The finished TDRObj is stored in S4Vectors::metadata(x)$tdr.obj.

If the assay is a DelayedMatrix (e.g., HDF5-backed), it is converted to a BPCells on-disk IterableMatrix and routed through .run_tdr_matrix() for efficient lazy access. If the assay is an in-memory matrix (e.g., dgCMatrix), the existing SCE backend path is used.
RunTDR(HDF5AnnData): Run the pipeline on an HDF5AnnData object

Builds a TDRObj from an HDF5-backed AnnData object. The expression matrix is converted to a BPCells on-disk directory for efficient lazy access, then the proven IterableMatrix pipeline is used for all downstream steps.

Metadata (obs) is read via anndataR; the expression matrix is read and converted via BPCells.
RunTDR(character): Run the pipeline directly from an h5ad file path

Reads metadata and gene/cell names from the h5ad file using rhdf5, opens the expression matrix with BPCells, and delegates to the internal IterableMatrix pipeline.

This method supports all H5AD format versions, including pre-0.8.0 files generated by older versions of Python anndata that are not supported by the anndataR package.
RunTDR(cytoset): Run the pipeline on a flowWorkspace cytoset

Builds a TDRObj from a cytoset object (one FCS sample per cytoframe) and executes the full pipeline.
RunTDR(flowSet): Run the pipeline on a flowCore flowSet

Builds a TDRObj from a flowSet object (one FCS sample per flowFrame) and executes the full pipeline. Requires only flowCore (not flowWorkspace).
RunTDR(dgCMatrix): Run the pipeline on a sparse matrix (dgCMatrix)

Builds a TDRObj from a dgCMatrix and per-cell metadata, then executes the full pipeline.
RunTDR(DelayedMatrix): Run the pipeline on a DelayedMatrix

Converts the DelayedMatrix to a BPCells on-disk IterableMatrix for efficient lazy access, then delegates to the proven IterableMatrix pipeline via .run_tdr_matrix().
RunTDR(IterableMatrix): Run the pipeline on a BPCells IterableMatrix