Run scTenifoldNet network comparison

Usage

RunscTenifoldNet(
  object,
  y = NULL,
  group.by = NULL,
  condition1 = NULL,
  condition2 = NULL,
  assay = NULL,
  layer = "counts",
  features = NULL,
  qc = TRUE,
  qc_min_library_size = 1000,
  qc_remove_outlier_cells = TRUE,
  qc_min_pct = 0.05,
  qc_max_mt_ratio = 0.1,
  nc_nNet = 10,
  nc_nCells = 500,
  nc_nComp = 3,
  nc_symmetric = FALSE,
  nc_scaleScores = TRUE,
  nc_q = 0.05,
  td_K = 3,
  td_nDecimal = 1,
  td_maxIter = 1000,
  td_maxError = 1e-05,
  ma_nDim = 30,
  cores = 1,
  store_networks = TRUE,
  store_manifold = TRUE,
  tool_name = "scTenifoldNet",
  verbose = TRUE
)

Arguments

object: A Seurat object or a raw count matrix with genes in rows and cells in columns.
y: A second raw count matrix. Required when object is a matrix and ignored when object is a Seurat object.
group.by: Metadata column used to split a Seurat object into the two conditions being compared.
condition1, condition2: Condition labels from group.by. If omitted, the first two group levels are used.
assay, layer: Assay and layer used as the count matrix when object is a Seurat object.
features: Optional genes to retain before running the comparison.
qc: Whether to apply scTenifoldNet-style quality control.
qc_min_library_size, qc_remove_outlier_cells, qc_min_pct, qc_max_mt_ratio: Quality-control parameters forwarded to scTenifoldNet::scQC().
nc_nNet, nc_nCells, nc_nComp, nc_symmetric, nc_scaleScores, nc_q: Network construction parameters forwarded to scTenifoldNet::makeNetworks().
td_K, td_nDecimal, td_maxIter, td_maxError: Tensor decomposition parameters forwarded to scTenifoldNet::tensorDecomposition().
ma_nDim: Manifold-alignment dimension forwarded to scTenifoldNet::manifoldAlignment().
cores: Number of cores forwarded to scTenifoldNet::scTenifoldNet().
store_networks: Whether to keep tensor networks in the stored result when object is a Seurat object.
store_manifold: Whether to keep manifold-alignment coordinates in the stored result when object is a Seurat object.
tool_name: Name of the object@tools entry when object is a Seurat object.
verbose: Whether to print the message. Default is TRUE.

Value

A scTenifoldNet result list for matrix input, or a Seurat object with results stored in object@tools[[tool_name]].

Examples

data(pancreas_sub)
counts <- GetAssayData5(pancreas_sub, assay = "RNA", layer = "counts")
detected <- names(sort(Matrix::rowSums(counts > 0), decreasing = TRUE))
features_use <- head(detected, 300)

pancreas_sub <- RunscTenifoldNet(
  pancreas_sub,
  group.by = "CellType",
  condition1 = "Ductal",
  condition2 = "Endocrine",
  features = features_use,
  qc = FALSE,
  nc_nNet = 3,
  nc_nCells = 200,
  td_maxIter = 200,
  ma_nDim = 2,
  store_networks = FALSE,
  store_manifold = TRUE
)
#> ℹ [2026-11-12 06:08:56] Run scTenifoldNet comparison using upstream implementation
#> 
#> ── scTenifoldNet Pipeline ──────────────────────────────────────────────────────
#> ℹ Step 2/6: CPM normalization
#> ℹ [X] CPM normalization: 300 genes x 253 cells
#> ✔ [X] CPM normalization complete
#> ℹ [Y] CPM normalization: 300 genes x 355 cells
#> ✔ [Y] CPM normalization complete
#> ℹ Shared genes: 300
#> ℹ Step 3/6: Building gene regulatory networks
#> ℹ [X] Building 3 gene regulatory networks (200 cells each)
#> ✔ [X] Network construction complete: 3 networks
#> ℹ [Y] Building 3 gene regulatory networks (200 cells each)
#> ✔ [Y] Network construction complete: 3 networks
#> ℹ Step 4/6: Tensor decomposition
#> ℹ [X] Tensor: 300 x 300 x 3 (K=3)
#> ✔ [X] CP decomposition complete (norm explained: 26.3%)
#> ℹ [Y] Tensor: 300 x 300 x 3 (K=3)
#> ✔ [Y] CP decomposition complete (norm explained: 24.7%)
#> ℹ Step 5/6: Manifold alignment
#> ℹ Manifold alignment: 300 shared genes, d=2
#> ✔ Manifold alignment complete: 2 dimensions
#> ℹ Step 6/6: Differential regulation analysis
#> ℹ Computing distances for 300 genes
#> ✔ Differential regulation complete: 2/300 significant genes (FDR < 0.05)
#> ✔ scTenifoldNet pipeline complete
#> ✔ [2026-11-12 06:08:57] scTenifoldNet results stored in `object@tools[[scTenifoldNet]]`

dr <- pancreas_sub@tools$scTenifoldNet$diffRegulation
head(dr)
#>         gene    distance        Z        FC      p.value       p.adj
#> 25  Hsp90ab1 0.004510967 1.690887 20.524738 5.886555e-06 0.001765966
#> 26      Rps2 0.004049140 1.605500 16.537272 4.770310e-05 0.007155465
#> 125    Hspa8 0.003194686 1.424532 10.294248 1.334455e-03 0.133445509
#> 177 Hsp90aa1 0.003058008 1.392084  9.432257 2.132018e-03 0.144242440
#> 47      Ftl1 0.003022130 1.383375  9.212227 2.404041e-03 0.144242440
#> 155   Sec61b 0.002831187 1.335586  8.084916 4.463519e-03 0.223175954

scTenifoldNetPlot(pancreas_sub, plot_type = "effect")