Run Scanorama on Seurat's Assay5 object through IntegrateLayers — ScanoramaIntegration • SeuratIntegrate

A wrapper to run Scanorama on multi-layered Seurat V5 object. Requires a conda environment with scanorama and necessary dependencies

Usage

ScanoramaIntegration(
  object,
  orig,
  layers = NULL,
  features = NULL,
  scale.layer = "scale.data",
  conda_env = NULL,
  new.reduction = "integrated.scanorama",
  reduction.key = "scanorama_",
  reconstructed.assay = "scanorama.reconstructed",
  ncores = 1L,
  ndims.out = 100L,
  return_dense = TRUE,
  batch_size = 5000,
  approx = TRUE,
  sigma = 15L,
  alpha = 0.1,
  knn = 20L,
  hvg.scanorama = NULL,
  union.features = FALSE,
  sketch = FALSE,
  sketch_method = c("geosketch", "uniform"),
  sketch_max = 10000,
  seed.use = 42L,
  verbose = TRUE,
  ...
)

Arguments

object: A Seurat object (or an Assay5 object if not called by IntegrateLayers)
orig: DimReduc object. Not to be set directly when called with IntegrateLayers, use orig.reduction argument instead
layers: Name of the layers to use in the integration
features: Vector of feature names to input to the integration method. When features = NULL (default), the VariableFeatures are used. To pass all features, use the output of Features()
scale.layer: Name of the scaled layer in Assay
conda_env: Path to conda environment to run scanorama (should also contain the scipy python module). By default, uses the conda environment registered for scanorama in the conda environment manager
new.reduction: Name to store resulting dimensional reduction object. NULL or FALSE disables the dimension reduction computation
reduction.key: Key for resulting dimensional reduction object. Ignored if reduction.key is NULL or FALSE
reconstructed.assay: Name for the assay containing the corrected expression matrix of dimension #features x #cells. NULL or FALSE disables the corrected expression matrix computation
ncores: Number of parallel threads to create through blas_set_num_threads. Pointless if BLAS is not supporting multithreaded
ndims.out: Number of dimensions for new.reduction output. Scanorama specific argument
return_dense: Whether scanorama returns numpy.ndarray matrices instead of scipy.sparse.csr_matrix. Scanorama specific argument
batch_size: Used in the alignment vector computation. Higer values increase memory burden.Scanorama specific argument
approx: Whether scanorama approximates nearest neighbors computation. Scanorama specific argument
sigma: Correction smoothing parameter on gaussian kernel. Scanorama specific argument
alpha: Minimum alignment score. Scanorama specific argument
knn: Number of nearest neighbours used for matching. Scanorama specific argument
hvg.scanorama: A positive integer to turn scanorama's internal HVG selection on. Disabled by default. Scanorama specific argument
union.features: By default, scanorama uses the intersection of features to perform integration. Set this parameter to TRUE to use the union. Discouraged. Scanorama specific argument
sketch: Turns on sketching-based acceleration by first downsampling the datasets. See Hie et al., Cell Systems (2019). Disabled by default. Ignored when reconstructed.assay is enabled. Scanorama specific argument
sketch_method: A skething method to apply to the data. Either 'geosketch' (default) or 'uniform'. Ignored when reconstructed.assay is enabled or when sketch is FALSE. Scanorama specific argument
sketch_max: Downsampling cutoff. Ignored when sketching disabled. Scanorama specific argument
seed.use: An integer to generate reproducible outputs. Set seed.use = NULL to disable
verbose: Print messages. Set to FALSE to disable
...: Ignored

Value

A list containing at least one of:

a new DimReduc of name reduction.name (key set to reduction.key) with corrected embeddings matrix of ndims.out.
a new Assay of name reconstructed.assay with corrected counts for each features (or less if hvg is set to a lower integer).

When called via IntegrateLayers, a Seurat object with the new reduction and/or assay is returned

Note

This function requires the Scanorama package to be installed (along with scipy)

References

Hie, B., Bryson, B. & Berger, B. Efficient integration of heterogeneous single-cell transcriptomes using Scanorama. Nat Biotechnol 37, 685–691 (2019). DOI

Examples

if (FALSE) { # \dontrun{
# Preprocessing
obj <- SeuratData::LoadData("pbmcsca")
obj[["RNA"]] <- split(obj[["RNA"]], f = obj$Method)
obj <- NormalizeData(obj)
obj <- FindVariableFeatures(obj)
obj <- ScaleData(obj)
obj <- RunPCA(obj)

# After preprocessing, we integrate layers:
obj <- IntegrateLayers(object = obj, method = ScanoramaIntegration)

# To disable feature expression matrix correction:
obj <- IntegrateLayers(object = obj, method = ScanoramaIntegration,
                       reconstructed.assay = NULL)
} # }