The WNN integration function

The WNN integration function

Usage

WNN_integrate(
  srt_merge = NULL,
  batch = NULL,
  append = TRUE,
  srt_list = NULL,
  assay = NULL,
  do_normalization = NULL,
  normalization_method = "LogNormalize",
  do_HVF_finding = TRUE,
  HVF_source = "separate",
  HVF_method = "vst",
  nHVF = 2000,
  HVF_min_intersection = 1,
  HVF = NULL,
  do_scaling = TRUE,
  vars_to_regress = NULL,
  regression_model = "linear",
  scale_within_batch = FALSE,
  linear_reduction = "pca",
  linear_reduction_dims = 50,
  linear_reduction_dims_use = NULL,
  linear_reduction_params = list(),
  force_linear_reduction = FALSE,
  nonlinear_reduction = "umap",
  nonlinear_reduction_dims = c(2, 3),
  nonlinear_reduction_params = list(),
  force_nonlinear_reduction = TRUE,
  neighbor_metric = "euclidean",
  neighbor_k = 20L,
  cluster_algorithm = "louvain",
  cluster_resolution = 0.6,
  verbose = TRUE,
  seed = 11
)

Arguments

srt_merge

A merged `Seurat` object that includes the batch information.

batch

A character string specifying the batch variable name.

append

Whether the integrated data will be appended to the original Seurat object (srt_merge). Default is TRUE.

srt_list

A list of Seurat objects to be checked and preprocessed.

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.

do_normalization

Whether data normalization should be performed. Default is TRUE.

normalization_method

The normalization method to be used. Possible values are "LogNormalize", "SCT", "TFIDF", and "scran". Default is "LogNormalize".

do_HVF_finding

Whether to perform high variable feature finding. If TRUE, the function will force to find the highly variable features (HVF) using the specified HVF method.

HVF_source

The source of highly variable features. Possible values are "global" and "separate". Default is "separate".

HVF_method

The method to use for finding highly variable features. Options are "vst", "mvp", "disp", or "scran". Default is "vst".

nHVF

The number of highly variable features to select. If NULL, all highly variable features will be used. Default is 2000.

HVF_min_intersection

The feature needs to be present in batches for a minimum number of times in order to be considered as highly variable. Default is 1.

HVF

A vector of feature names to use as highly variable features. If NULL, the function will use the highly variable features identified by the HVF method.

do_scaling

Whether to perform scaling. If TRUE, the function will force to scale the data using the Seurat::ScaleData function.

vars_to_regress

A vector of variable names to include as additional regression variables. Default is NULL.

regression_model

The regression model to use for scaling. Options are "linear", "poisson", or "negativebinomial". Default is "linear".

scale_within_batch

Whether to scale data within each batch. Only valid when the integration_method is one of "Uncorrected", "Seurat", "MNN", "Harmony", "BBKNN", "CSS", "ComBat".

linear_reduction

The linear dimensionality reduction method to use. Options are "pca", "svd", "ica", "nmf", "mds", or "glmpca". Default is "pca".

linear_reduction_dims

The number of dimensions to keep after linear dimensionality reduction. Default is 50.

linear_reduction_dims_use

The dimensions to use for downstream analysis. If NULL, estimated dimensions stored in the linear reduction will be used when available; otherwise, the first up to 50 dimensions will be used as a fallback.

linear_reduction_params

A list of parameters to pass to the linear dimensionality reduction method.

force_linear_reduction

Whether to force linear dimensionality reduction even if the specified reduction is already present in the Seurat object.

nonlinear_reduction

The nonlinear dimensionality reduction method to use. Options are "umap", "umap-naive", "tsne", "dm", "phate", "pacmap", "trimap", "largevis", or "fr". Default is "umap".

nonlinear_reduction_dims

The number of dimensions to keep after nonlinear dimensionality reduction. If a vector is provided, different numbers of dimensions can be specified for each method. Default is c(2, 3).

nonlinear_reduction_params

A list of parameters to pass to the nonlinear dimensionality reduction method.

force_nonlinear_reduction

Whether to force nonlinear dimensionality reduction even if the specified reduction is already present in the Seurat object. Default is TRUE.

neighbor_metric

The distance metric to use for finding neighbors. Options are "euclidean", "cosine", "manhattan", or "hamming". Default is "euclidean".

neighbor_k

The number of nearest neighbors to use for finding neighbors. Default is 20.

cluster_algorithm

The clustering algorithm to use. Options are "louvain", "slm", or "leiden". Default is "louvain".

cluster_resolution

The resolution parameter to use for clustering. Larger values result in fewer clusters. Default is 0.6.

verbose

Whether to print the message. Default is TRUE.

seed

Random seed for reproducibility. Default is 11.

Examples

data("pbmcmultiome_sub", package = "scop")
pbmcmultiome_sub$batch <- rep(c("batch1", "batch2"), length.out = ncol(pbmcmultiome_sub))
pbmcmultiome_sub <- WNN_integrate(
  srt_merge = pbmcmultiome_sub,
  batch = "batch",
  linear_reduction_dims = 20,
  linear_reduction_dims_use = 1:10
)
#> ℹ [2026-05-14 07:52:24] Start standard processing workflow...
#> ℹ [2026-05-14 07:52:24] Auto preprocess assays: "RNA" and "peaks"
#> ℹ [2026-05-14 07:52:24] Start standard processing workflow...
#> ℹ [2026-05-14 07:52:24] Checking a list of <Seurat>...
#> ! [2026-05-14 07:52:24] Data 1/1 of the `srt_list` is "unknown"
#> ℹ [2026-05-14 07:52:24] Perform `NormalizeData()` with `normalization.method = 'LogNormalize'` on 1/1 of `srt_list`...
#> ℹ [2026-05-14 07:52:26] Perform `Seurat::FindVariableFeatures()` on 1/1 of `srt_list`...
#> Warning: pseudoinverse used at -2.3979
#> Warning: neighborhood radius 0.30103
#> Warning: reciprocal condition number  1.2589e-15
#> ℹ [2026-05-14 07:52:27] Use the separate HVF from `srt_list`
#> ℹ [2026-05-14 07:52:27] Number of available HVF: 2000
#> ℹ [2026-05-14 07:52:27] Finished check
#> ℹ [2026-05-14 07:52:27] Perform `Seurat::ScaleData()`
#> ℹ [2026-05-14 07:52:27] Perform pca linear dimension reduction
#> ℹ [2026-05-14 07:52:28] Perform `Seurat::FindClusters()` with `cluster_algorithm = 'louvain'` and `cluster_resolution = 0.6`
#> ℹ [2026-05-14 07:52:28] Reorder clusters...
#> ℹ [2026-05-14 07:52:28] Skip `log1p()` because `layer = data` is not "counts"
#> ℹ [2026-05-14 07:52:28] Perform umap nonlinear dimension reduction
#> ℹ [2026-05-14 07:52:28] Perform umap nonlinear dimension reduction using RNApca (1:10)
#> ℹ [2026-05-14 07:52:34] Perform umap nonlinear dimension reduction using RNApca (1:10)
#> ✔ [2026-05-14 07:52:40] Standard processing workflow completed
#> ℹ [2026-05-14 07:52:40] Start standard processing workflow...
#> ℹ [2026-05-14 07:52:40] Checking a list of <Seurat>...
#> ! [2026-05-14 07:52:40] Data 1/1 of the `srt_list` is "raw_counts"
#> ℹ [2026-05-14 07:52:40] Perform `RunTFIDF()` on 1/1 of `srt_list`...
#> ℹ [2026-05-14 07:52:40] Perform `FindTopFeatures()` on 1/1 of `srt_list`...
#> ℹ [2026-05-14 07:52:41] Use the separate HVF from `srt_list`
#> ℹ [2026-05-14 07:52:41] Number of available HVF: 11413
#> ℹ [2026-05-14 07:52:41] Finished check
#> ℹ [2026-05-14 07:52:41] `normalization_method` is TFIDF. Use lsi workflow
#> ℹ [2026-05-14 07:52:41] Perform svd linear dimension reduction
#> Running SVD
#> Scaling cell embeddings
#> ℹ [2026-05-14 07:52:42] Perform `Seurat::FindClusters()` with `cluster_algorithm = 'louvain'` and `cluster_resolution = 0.6`
#> ℹ [2026-05-14 07:52:42] Reorder clusters...
#> ℹ [2026-05-14 07:52:42] Skip `log1p()` because `layer = data` is not "counts"
#> ℹ [2026-05-14 07:52:42] Perform umap nonlinear dimension reduction
#> ℹ [2026-05-14 07:52:42] Perform umap nonlinear dimension reduction using ATACsvd (1:10)
#> ℹ [2026-05-14 07:52:48] Perform umap nonlinear dimension reduction using ATACsvd (1:10)
#> ✔ [2026-05-14 07:52:53] Standard processing workflow completed
#> ℹ [2026-05-14 07:52:53] Adjust neighbor k from 20 to 20 for small-sample WNN graph construction
#> ℹ [2026-05-14 07:52:53] Adjust WNN knn.range to 80 for small-sample graph construction
#> ℹ [2026-05-14 07:52:53] Perform WNN integration using RNApca and ATAClsi
#> Calculating cell-specific modality weights
#> Finding 20 nearest neighbors for each modality.
#> Calculating kernel bandwidths
#> Finding multimodal neighbors
#> Constructing multimodal KNN graph
#> Constructing multimodal SNN graph
#> ℹ [2026-05-14 07:52:55] Adjust neighbor k from 20 to 20 for small-sample clustering
#> ℹ [2026-05-14 07:52:55] Perform `Seurat::FindClusters()` with "louvain"
#> ℹ [2026-05-14 07:52:55] Reorder clusters...
#> ℹ [2026-05-14 07:52:55] Skip `log1p()` because `layer = data` is not "counts"
#> ℹ [2026-05-14 07:52:55] Perform umap nonlinear dimension reduction using WNN
#> ℹ [2026-05-14 07:53:01] Perform umap nonlinear dimension reduction using WNN
#> Warning: Key ‘RNApcaUMAP2D_’ taken, using ‘rnaumap2d_’ instead
#> Warning: Key ‘RNApcaUMAP3D_’ taken, using ‘rnaumap3d_’ instead