TDRObj: S4 class for tinydenseR analysis objects

The core data structure for tinydenseR landmark-based analysis. Contains expression data references, metadata, dimensionality reduction results, graph structure, and differential expression results.

Usage

# S4 method for class 'TDRObj'
x$name

# S4 method for class 'TDRObj'
x$name <- value

# S4 method for class 'TDRObj'
names(x)

# S4 method for class 'TDRObj'
show(object)

Arguments

x: A TDRObj object.
name: A character string naming the slot to access.
value: The value to assign to the slot.
object: A TDRObj object (used in show method).

Slots

cells

list. Named list of per-sample file paths to expression matrices.

metadata

data.frame. Sample-level metadata.

config

list. Run parameters: key, sampling, assay.type, markers, n.threads.

integration

list. Trained projection models and batch variables (harmony.var, harmony.obj, symphony.obj, umap.model).

assay

list. Landmark expression layers (L x features matrices): raw (raw counts), expr (normalized/log expression), scaled (Z-scored).

landmark.embed

list. Landmark-space coordinate matrices; each entry has a $coord slot. Contains pca, le, and umap sub-lists.

landmark.annot

list. Per-landmark categorical annotations (factor, length L). Contains clustering and celltyping sub-lists, each with an $ids factor.

graphs

list. Landmark-landmark connectivity matrices: adj.matrix, snn, fgraph.

density

list. Fuzzy density analytics populated by get.map. Contains five sub-elements:

raw: L × N matrix of pre-normalization fuzzy density sums.
norm: L × N matrix after size-factor normalization: norm = t(t(raw) / size.factors).
log.norm: L × N matrix: log2(norm + 0.5).
size.factors: Named numeric(N): $n_j / \bar{n}$, guaranteed to have mean 1.
composition: List of clustering/celltyping cell count and percentage matrices (samples × clusters).

Access via get.density or the $ accessor.

sample.embed

list. Sample-level embeddings (N x k matrices), each with $coord. Contains pca, traj, and pepc sub-lists.

cellmap

list. Per-cell, per-sample data in unified structure: clustering$ids, celltyping$ids (named per-sample lists with optional named solutions), nearest.lm, fuzzy.graphs. Each sample entry is either an in-memory R object or an attributed path string for on-disk cache.

results

list. All statistical outputs: lm, pb, marker, spec, nmf, pls, features.

TDRObj: S4 class for tinydenseR analysis objects

Usage

Arguments

Slots

See also