harmony (version 1.2.1)

RunHarmony.default: This is the primary harmony interface.

Description

Use this generic with a cell embeddings matrix, a metadata table and a categorical covariate to run the Harmony algorithm directly on cell embedding matrix.

Usage

# S3 method for default
RunHarmony(
  data_mat,
  meta_data,
  vars_use,
  theta = NULL,
  sigma = 0.1,
  lambda = 1,
  nclust = NULL,
  max_iter = 10,
  early_stop = TRUE,
  ncores = 1,
  plot_convergence = FALSE,
  return_object = FALSE,
  verbose = TRUE,
  .options = harmony_options(),
  ...
)

Value

By default, matrix with corrected PCA embeddings. If return_object is TRUE, returns the full Harmony object (R6 reference class type).

Arguments

data_mat

Matrix of cell embeddings. Cells can be rows or columns and will be inferred by the rows of meta_data.

meta_data

Either (1) Dataframe with variables to integrate or (2) vector with labels.

vars_use

If meta_data is dataframe, this defined which variable(s) to remove (character vector).

theta

Diversity clustering penalty parameter. Specify for each variable in vars_use Default theta=2. theta=0 does not encourage any diversity. Larger values of theta result in more diverse clusters.

sigma

Width of soft kmeans clusters. Default sigma=0.1. Sigma scales the distance from a cell to cluster centroids. Larger values of sigma result in cells assigned to more clusters. Smaller values of sigma make soft kmeans cluster approach hard clustering.

lambda

Ridge regression penalty. Default lambda=1. Bigger values protect against over correction. If several covariates are specified, then lambda can also be a vector which needs to be equal length with the number of variables to be corrected. In this scenario, each covariate level group will be assigned the scalars specified by the user. If set to NULL, harmony will start lambda estimation mode to determine lambdas automatically and try to minimize overcorrection (Use with caution still in beta testing).

nclust

Number of clusters in model. nclust=1 equivalent to simple linear regression.

max_iter

Maximum number of rounds to run Harmony. One round of Harmony involves one clustering and one correction step.

early_stop

Enable early stopping for harmony. The harmonization process will stop when the change of objective function between corrections drops below 1e-4

ncores

Number of processors to be used for math operations when optimized BLAS is available. If BLAS is not supporting multithreaded then this option has no effect. By default, ncore=1 which runs as a single-threaded process. Although Harmony supports multiple cores, it is not optimized for multithreading. Increase this number for large datasets iff single-core performance is not adequate.

plot_convergence

Whether to print the convergence plot of the clustering objective function. TRUE to plot, FALSE to suppress. This can be useful for debugging.

return_object

(Advanced Usage) Whether to return the Harmony object or only the corrected PCA embeddings.

verbose

Whether to print progress messages. TRUE to print, FALSE to suppress.

.options

Setting advanced parameters of RunHarmony. This must be the result from a call to `harmony_options`. See ?`harmony_options` for parameters not listed above and more details.

...

other parameters that are not part of the API

See Also

Other RunHarmony: RunHarmony.Seurat(), RunHarmony.SingleCellExperiment(), RunHarmony()

Examples

## By default, Harmony inputs a cell embedding matrix
if (FALSE) {
harmony_embeddings <- RunHarmony(cell_embeddings, meta_data, 'dataset')
}

## If PCA is the input, the PCs need to be scaled
data(cell_lines_small)
pca_matrix <- cell_lines_small$scaled_pcs
meta_data <- cell_lines_small$meta_data
harmony_embeddings <- RunHarmony(pca_matrix, meta_data, 'dataset')

## Output is a matrix of corrected PC embeddings
dim(harmony_embeddings)
harmony_embeddings[seq_len(5), seq_len(5)]

## Finally, we can return an object with all the underlying data structures
harmony_object <- RunHarmony(pca_matrix, meta_data, 'dataset', return_object=TRUE)
dim(harmony_object$Y) ## cluster centroids
dim(harmony_object$R) ## soft cluster assignment
dim(harmony_object$Z_corr) ## corrected PCA embeddings
head(harmony_object$O) ## batch by cluster co-occurence matrix