Harmony fix

module-cheatsheets/scRNA-seq-advanced-cheatsheet.md

@@ -77,7 +77,8 @@ Read the [`harmony` package documentation](https://rdrr.io/github/immunogenomics
 |----------------------|--------------------|---------------------|---------------|
 | `batchelor`| [`MultiBatchPCA()`](https://rdrr.io/bioc/batchelor/man/multiBatchPCA.html)| Multi-batch PCA | Perform PCA across multiple gene expression matrices, weighted by batch size |
 | `batchelor`| [`fastMNN()`](https://rdrr.io/bioc/batchelor/man/fastMNN.html)| Fast mutual nearest neighbors correction | Perform integration on an SCE object with mutual nearest neighbors using the `fastMNN` algorithm, returning an SCE object with batch-corrected principal components |
-| `harmony`| [`HarmonyMatrix()`](https://rdrr.io/github/immunogenomics/harmony/man/HarmonyMatrix.html)| Perform `harmony` integration on a matrix | Perform integration with `harmony` on either a matrix of principle components or gene expression, returning a matrix of batch-corrected principal components  |
+| `harmony`| [`RunHarmony()`](https://rdrr.io/github/immunogenomics/harmony/man/RunHarmony.html)| Run the `harmony` algorithm | Perform integration with the `harmony` algorithm on a matrix of single-cell genomics cell embeddings, returning a matrix of batch-corrected principal components  |
+
 
 ## `SingleR`
 

scRNA-seq-advanced/02-dataset_integration.Rmd

@@ -739,9 +739,8 @@ Like `fastMNN`, `harmony` performs integration on a merged PCA matrix.
 However, unlike `fastMNN`, `harmony` does not "back-calculate" corrected expression from the corrected PCA matrix and it only returns the corrected PCA matrix itself.
 For input, `harmony` needs a couple pieces of information:
 
-- First, `harmony` can either take a matrix of normalized expression values, from which it will calculate a batch-weighted PCA matrix to integrate, or it can take a batch-weighted PCA matrix directly to perform integration.
+- First, `harmony` takes batch-weighted PCA matrix directly to perform integration.
 Since we already calculated a batch-weighted PCA matrix (our `merged_PCA` reduced dimension), we'll provide this information directly.
-  - We will need to specify the additional argument `do_pca=FALSE` to tell `harmony` that the input matrix we provided already is a PCA matrix.
 - Second, we need to tell `harmony` about the covariates to use - `sample` and `patient`.
 To do this, we provide two arguments:
   - `meta_data`, a data frame that contains covariates across samples.
@@ -752,11 +751,9 @@ To do this, we provide two arguments:
 Let's go!
 
 ```{r run harmony, live = TRUE}
-# integrate with harmony, setting the argument `do_pca = FALSE`
-#  since we are providing a PCA matrix directly
-harmony_pca <- harmony::HarmonyMatrix(
+# Run harmony integration
+harmony_pca <- harmony::RunHarmony(
   data_mat = reducedDim(merged_sce, "merged_PCA"),
-  do_pca = FALSE,
   meta_data = colData(merged_sce),
   vars_use = c("sample", "patient")
 )