Run scTenifoldKnk in-silico knockout analysis

Usage

RunscTenifoldKnk(
  srt,
  gKO,
  assay = NULL,
  layer = "counts",
  features = NULL,
  qc = TRUE,
  qc_mt_threshold = 0.1,
  qc_min_library_size = 1000,
  qc_min_cells = 25,
  nc_lambda = 0,
  nc_nNet = 10,
  nc_nCells = 500,
  nc_nComp = 3,
  nc_scaleScores = TRUE,
  nc_symmetric = FALSE,
  nc_q = 0.9,
  td_K = 3,
  td_maxIter = 1000,
  td_maxError = 1e-05,
  td_nDecimal = 3,
  ma_nDim = 2,
  cores = 1,
  backend = c("cpp", "scTenifoldKnk"),
  store_networks = TRUE,
  store_manifold = TRUE,
  tool_name = "scTenifoldKnk",
  verbose = TRUE
)

Arguments

srt: A Seurat object.
gKO: Gene symbol or symbols to knock out. All genes must be present after optional feature and QC filtering.
assay: Which assay to use. If NULL, the default assay of the Seurat object will be used. When the object also contains ChromatinAssay, the default assay and additional ChromatinAssay will be preprocessed sequentially.
layer: Assay layer used as the count matrix.
features: Optional genes to retain before running network construction. If supplied, gKO is always retained when present in the input assay.
qc: Whether to apply scTenifoldKnk-style quality control.
qc_mt_threshold: Maximum mitochondrial read fraction per cell.
qc_min_library_size: Minimum library size per cell.
qc_min_cells: Minimum number of expressing cells required per gene.
nc_lambda, nc_nNet, nc_nCells, nc_nComp, nc_scaleScores, nc_symmetric, nc_q: Network construction parameters forwarded to scTenifoldNet::makeNetworks().
td_K, td_maxIter, td_maxError, td_nDecimal: Tensor decomposition parameters forwarded to scTenifoldNet::tensorDecomposition().
ma_nDim: Manifold-alignment dimension forwarded to scTenifoldNet::manifoldAlignment().
cores: Number of cores used by native network-construction workers and forwarded to downstream linear algebra where applicable.
backend: cpp is a native equivalent covariance-based network construction, direct sparse network assembly, controlled per-gene eigensolver parallelism, and helpers for tensor decomposition, manifold matrix construction, directionality, and distance calculation. scTenifoldKnk calls scTenifoldKnk::scTenifoldKnk() directly for comparison.
store_networks: Whether to keep WT/KO tensor networks in srt@tools.
store_manifold: Whether to keep manifold-alignment coordinates in srt@tools.
tool_name: Name of the srt@tools entry.
verbose: Whether to print the message. Default is TRUE.

Value

A Seurat object with scTenifoldKnk results stored in srt@tools[[tool_name]].

Examples

data(pancreas_sub)
gene_use <- "Pdx1"
counts <- GetAssayData5(pancreas_sub, assay = "RNA", layer = "counts")
detected <- names(sort(Matrix::rowSums(counts > 0), decreasing = TRUE))
features_use <- unique(c(gene_use, head(detected, 300)))

pancreas_sub <- RunscTenifoldKnk(
  pancreas_sub,
  gKO = gene_use,
  features = features_use,
  qc = FALSE,
  nc_nNet = 3,
  nc_nCells = 200,
  td_maxIter = 200,
  store_networks = FALSE,
  store_manifold = TRUE
)
#> ℹ [2026-05-14 07:37:22] Run scTenifoldKnk knockout for "Pdx1" using "cpp" backend
#> ℹ [2026-05-14 07:37:27] Construct scTenifoldNet network ensemble
#> ℹ [2026-05-14 07:37:27] Denoise network ensemble with tensor decomposition
#> ℹ [2026-05-14 07:37:27] Denoise network ensemble with tensor decomposition  ■■ …
#> ℹ [2026-05-14 07:37:27] Align WT and KO network manifolds
#> ✔ [2026-05-14 07:37:27] scTenifoldKnk results stored in `srt@tools[[scTenifoldKnk]]`

dr <- pancreas_sub@tools$scTenifoldKnk$diffRegulation
head(dr)
#>     gene     distance        Z          FC      p.value        p.adj
#> 1   Pdx1 9.304567e-04 4.568187 12059.91897 0.000000e+00 0.000000e+00
#> 2   Gnas 4.887597e-05 2.057742    33.27690 7.992633e-09 1.202891e-06
#> 3   Ssr2 4.386627e-05 1.985077    26.80486 2.250708e-07 2.258210e-05
#> 4   Ssr4 4.132486e-05 1.945477    23.78893 1.074990e-06 8.089300e-05
#> 5 Sec61b 4.074745e-05 1.936192    23.12879 1.515051e-06 9.120607e-05
#> 6   Dad1 3.408756e-05 1.820118    16.18616 5.741213e-05 2.880175e-03

scTenifoldKnkPlot(pancreas_sub, plot_type = "effect")