Skip to contents

Find Clusters in Feature Data

find_clusters.Rd

Performs cluster analysis on feature data using shared nearest neighbor (SNN) clustering.

Internal functions adapted from the Seurat package for neighbor finding and cluster analysis. These functions are modified versions of the original Seurat code.

Usage

find_clusters(x, ...)

# Default S3 method
find_clusters(
  x,
  metadata_cols,
  k.param = 20,
  prune.SNN = 1/15,
  n.pcs = 20,
  resolution = 0.2,
  n.start = 10,
  verbose = TRUE,
  ...
)

# S3 method for class 'Sap'
find_clusters(
  x,
  segment_type = c("motifs", "syllables", "segments"),
  data_type = c("spectral_feature", "spectrogram"),
  label = NULL,
  k.param = 20,
  prune.SNN = 1/15,
  n.pcs = NULL,
  resolution = NULL,
  n.start = 10,
  verbose = TRUE,
  ...
)

Arguments

x

An object to analyze, either a data frame or SAP object

...

Additional arguments passed to specific methods

metadata_cols

For default method: Column indices for metadata

k.param

Number of nearest neighbors (default: 20)

prune.SNN

Pruning threshold for SNN graph (default: 1/15)

n.pcs

Number of principal components to use

resolution

Resolution parameter for clustering

n.start

Number of random starts (default: 10)

verbose

Whether to print progress messages (default: TRUE)

segment_type

For SAP objects: Type of segments ('motifs', 'syllables', 'segments')

data_type

For SAP objects: Type of feature data ('spectral_feature', "spectrogram")

label

For SAP objects: Specific label to filter data

Value

For default method: Data frame with metadata columns and cluster assignments For SAP objects: Updated object with clustering results in features slot

Details

For feature data frames:

  • Separates metadata and feature columns

  • Finds nearest neighbors using PCA

  • Constructs SNN graph

  • Performs community detection

For SAP objects:

  • Supports multiple segment types

  • Optional label filtering

  • Stores results in features slot

  • Updates feature embeddings

The clustering approach is similar to that used in Seurat V3, implementing:

  • PCA-based neighbor finding

  • SNN graph construction

  • Louvain community detection

Original source: Seurat package (https://github.com/satijalab/seurat/blob/main/inst/CITATION)

These functions are based on methods developed in the following publications:

  • Hao Y, et al. (2023). Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nature Biotechnology. doi:10.1038/s41587-023-01767-y

  • Hao Y, Hao S, et al. (2021). Integrated analysis of multimodal single-cell data. Cell. doi:10.1016/j.cell.2021.04.048

  • Stuart T, Butler A, et al. (2019). Comprehensive Integration of Single-Cell Data. Cell. doi:10.1016/j.cell.2019.05.031

  • Butler A, et al. (2018). Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nature Biotechnology. doi:10.1038/nbt.4096

  • Satija R, et al. (2015). Spatial reconstruction of single-cell gene expression data. Nature Biotechnology. doi:10.1038/nbt.3192

See also

run_umap for visualization of clusters

Examples

if (FALSE) { # \dontrun{
# Basic clustering of feature data
clusters <- find_clusters(features,
                         metadata_cols = c(1:5))

# Clustering with custom parameters
clusters <- find_clusters(features,
                         metadata_cols = c(1:5),
                         k.param = 30,
                         resolution = 0.3)

# Cluster SAP object features
sap_obj <- find_clusters(sap_object,
                        segment_type = "motifs",
                        data_type = "spectral_feature")

# Label-specific clustering
sap_obj <- find_clusters(sap_object,
                        segment_type = "syllables",
                        label = "a",
                        resolution = 0.4)
} # }