5 Normalisation and dimensionality reduction
5.1 Just got back to 1 worker for this analysis seems to work much better
plan("multisession", workers = 1)5.2 Log normalise
cropped.obj <- NormalizeData(cropped.obj)
cropped.obj <- FindVariableFeatures(cropped.obj, selection.method = "vst")5.3 Identify the 20 most highly variable genes
t20 <- head(VariableFeatures(cropped.obj), 20)
t205.4 Scale data and run PCA
cropped.obj <- ScaleData(cropped.obj)
cropped.obj <- RunPCA(cropped.obj, features = VariableFeatures(object = cropped.obj))
print(cropped.obj[["pca"]], dims = 1:5, nfeatures = 5)5.5 Find neighbour cells (in PCA space, not real space)
cropped.obj <- FindNeighbors(cropped.obj, dims = 1:30)5.6 Run with default cluster params
cropped.obj <- FindClusters(cropped.obj)5.7 Run UMAP
cropped.obj <- RunUMAP(cropped.obj, dims = 1:30)5.8 Markers of each cluster
marks <- FindAllMarkers(cropped.obj)