Skip to content

Commit

Permalink
Merge pull request #800 from AlexsLemonade/sjspielman/692-734-scadvan…
Browse files Browse the repository at this point in the history 
…ced-updates

scAdvanced updates: dir_create and plotting
  • Loading branch information
@sjspielman
sjspielman authored Nov 5, 2024
2 parents 8718d22 + d954ed6 commit b5d6474
Show file tree
Hide file tree
Showing 4 changed files with 38 additions and 39 deletions.
8 changes: 3 additions & 5 deletions scRNA-seq-advanced/01-read_filter_normalize_scRNA.Rmd
View file
Original file line number Diff line number Diff line change
Expand Up @@ -93,9 +93,7 @@ mito_file <- file.path(ref_dir, "hs_mitochondrial_genes.tsv")
normalized_dir <- file.path(data_dir, "normalized")
# create the directory if it does not exist
if (!dir.exists(normalized_dir)) {
dir.create(normalized_dir, recursive = TRUE)
}
fs::dir_create(normalized_dir)
# output RDS file for normalized data
output_sce_file <- file.path(normalized_dir,
Expand Down Expand Up @@ -497,7 +495,7 @@ We can also visualize the UMAP results using the `plotReducedDim()` function.
scater::plotReducedDim(normalized_sce,
"UMAP",
# color by the most variable gene
colour_by = hv_genes[1])
color_by = hv_genes[1])
```


Expand Down Expand Up @@ -551,7 +549,7 @@ Here rather than the general `plotReducedDim()` function, we will use `plotUMAP(
```{r plot clusters, live=TRUE}
# plot UMAP with assigned clusters
scater::plotUMAP(normalized_sce,
colour_by = "nn_cluster")
color_by = "nn_cluster")
```

What do you see in these results?
Expand Down
48 changes: 26 additions & 22 deletions scRNA-seq-advanced/02-dataset_integration.Rmd
View file
Original file line number Diff line number Diff line change
Expand Up @@ -76,9 +76,7 @@ input_dir <- file.path("data", "rms", "processed")
output_dir <- file.path("data", "rms", "integrated")
# Create output directory if it doesn't exist
if (!(dir.exists(output_dir))) {
dir.create(output_dir)
}
fs::dir_create(output_dir)
# Define output file name for the integrated object
integrated_sce_file <- file.path(output_dir, "rms_integrated_subset.rds")
Expand Down Expand Up @@ -466,11 +464,15 @@ To see how this looks, let's look at the UMAP when calculated from individual sa
# Plot UMAP calculated from individual samples with separate scaling
scater::plotReducedDim(merged_sce,
dimred = "UMAP",
colour_by = "sample",
color_by = "sample",
point_size = 0.5,
point_alpha = 0.2) +
guides(colour = guide_legend(override.aes = list(size = 3, alpha = 1))) +
# Use a CVD-friendly color scheme
scale_color_brewer(palette = "Dark2") +
# Modify the legend key so its points are larger and easier to see
guides(color = guide_legend(override.aes = list(size = 3, alpha = 1))) +
ggtitle("UMAP calculated on each sample separately")
```


Expand Down Expand Up @@ -554,13 +556,12 @@ Now, let's see how this new `merged_UMAP` looks compared to the `UMAP` calculate
# UMAPs scaled together when calculated from the merged SCE
scater::plotReducedDim(merged_sce,
dimred = "merged_UMAP",
colour_by = "sample",
color_by = "sample",
# Some styling to help us see the points:
point_size = 0.5,
point_alpha = 0.2) +
# Modify the legend key so its points are larger and easier to see
guides(colour = guide_legend(override.aes = list(size = 3, alpha = 1))) +
# Add a plot title
scale_color_brewer(palette = "Dark2") +
guides(color = guide_legend(override.aes = list(size = 3, alpha = 1))) +
ggtitle("UMAP calculated on merged_sce")
```

Expand Down Expand Up @@ -633,13 +634,12 @@ scater::plotReducedDim(merged_sce,
# plot the fastMNN coordinates
dimred = "fastmnn_UMAP",
# color by sample
colour_by = "sample",
color_by = "sample",
# Some styling to help us see the points:
point_size = 0.5,
point_alpha = 0.2) +
# Modify legend so they key is larger and easier to see
guides(colour = guide_legend(override.aes = list(size = 3, alpha = 1))) +
# add plot title
scale_color_brewer(palette = "Dark2") +
guides(color = guide_legend(override.aes = list(size = 3, alpha = 1))) +
ggtitle("UMAP after integration with fastMNN")
```

Expand Down Expand Up @@ -679,10 +679,11 @@ Let's re-plot this UMAP to highlight cell types:
scater::plotReducedDim(merged_sce,
dimred = "fastmnn_UMAP",
# color by broad celltypes
colour_by = "celltype_broad",
color_by = "celltype_broad",
point_size = 0.5,
point_alpha = 0.2) +
guides(colour = guide_legend(override.aes = list(size = 3, alpha = 1))) +
scale_color_brewer(palette = "Dark2") +
guides(color = guide_legend(override.aes = list(size = 3, alpha = 1))) +
ggtitle("UMAP after integration with fastMNN")
```

Expand All @@ -695,12 +696,13 @@ One way we can see all the points a bit better is to facet the plot by sample, u
```{r plot fastmnn umap celltypes faceted}
scater::plotReducedDim(merged_sce,
dimred = "fastmnn_UMAP",
colour_by = "celltype_broad",
color_by = "celltype_broad",
point_size = 0.5,
point_alpha = 0.2,
# Allow for faceting by a variable using `other_fields`:
other_fields = "sample") +
guides(colour = guide_legend(override.aes = list(size = 3, alpha = 1))) +
scale_color_brewer(palette = "Dark2") +
guides(color = guide_legend(override.aes = list(size = 3, alpha = 1))) +
ggtitle("UMAP after integration with fastMNN") +
# Facet by sample
facet_wrap(vars(sample)) +
Expand Down Expand Up @@ -788,11 +790,12 @@ Let's see how the `harmony` UMAP, colored by sample, looks compared to the `fast
```{r plot harmony umap batches}
scater::plotReducedDim(merged_sce,
dimred = "harmony_UMAP",
colour_by = "sample",
color_by = "sample",
point_size = 0.5,
point_alpha = 0.2) +
ggtitle("UMAP after integration with harmony") +
guides(colour = guide_legend(override.aes = list(size = 3, alpha = 1)))
scale_color_brewer(palette = "Dark2") +
guides(color = guide_legend(override.aes = list(size = 3, alpha = 1))) +
ggtitle("UMAP after integration with harmony")
```

How do you think this `harmony` UMAP compares to that from `fastMNN` integration?
Expand All @@ -802,13 +805,14 @@ Let's see how this UMAP looks colored by cell type, and faceted for visibility:
```{r plot harmony umap celltypes}
scater::plotReducedDim(merged_sce,
dimred = "harmony_UMAP",
colour_by = "celltype_broad",
color_by = "celltype_broad",
point_size = 0.5,
point_alpha = 0.2,
# Specify variable for faceting
other_fields = "sample") +
scale_color_brewer(palette = "Dark2", na.value = "grey80") +
guides(color = guide_legend(override.aes = list(size = 3))) +
ggtitle("UMAP after integration with harmony") +
guides(colour = guide_legend(override.aes = list(size = 3))) +
facet_wrap(vars(sample))
```

Expand Down
17 changes: 8 additions & 9 deletions scRNA-seq-advanced/03-differential_expression.Rmd
View file
Original file line number Diff line number Diff line change
Expand Up @@ -79,9 +79,7 @@ sample_metadata_file <- file.path(data_dir,
# directory to store output
deseq_dir <- file.path("analysis", "rms", "deseq")
if(!dir.exists(deseq_dir)){
dir.create(deseq_dir, recursive = TRUE)
}
fs::dir_create(deseq_dir)
# results file to output from DE analysis
deseq_output_file <- file.path(deseq_dir,
Expand Down Expand Up @@ -208,7 +206,7 @@ In the chunk below we will start by taking a look at our integration results and
# UMAP of all samples, separating by diagnosis group
scater::plotReducedDim(integrated_sce,
dimred = "fastmnn_UMAP",
colour_by = "diagnosis_group",
color_by = "diagnosis_group",
point_size= 0.5,
point_alpha = 0.2)
```
Expand All @@ -230,7 +228,7 @@ The samples which could be further classified have a mix of `Tumor_Mesoderm`, `T
scater::plotReducedDim(integrated_sce,
dimred = "fastmnn_UMAP",
# color each point by cell type
colour_by = "celltype_broad",
color_by = "celltype_broad",
point_size= 0.5,
point_alpha = 0.4)
```
Expand All @@ -250,7 +248,7 @@ In the below plot we will color our cells by cell type while also using `facet_g
scater::plotReducedDim(integrated_sce,
dimred = "fastmnn_UMAP",
# color each point by cell type
colour_by = "celltype_broad",
color_by = "celltype_broad",
point_size= 0.5,
point_alpha = 0.4,
# tell scater to use diagnosis_group for plotting
Expand All @@ -277,6 +275,7 @@ ggplot(tumor_cells_df, aes(x = sample, fill = celltype_broad)) +
y = "Proportion of cells",
fill = "Cell type"
) +
scale_fill_brewer(palette = "Dark2") +
theme_bw() +
theme(axis.text.x = element_text(angle = 90, vjust = 0.5))+
# facet by diagnosis group
Expand Down Expand Up @@ -681,7 +680,7 @@ myoblast_combined_sce <- rms_sce[, which(rms_sce$celltype_broad == "Tumor_Myobla
# plot PTPRT (ENSG00000196090) expression in ARMS vs. ERMS
scater::plotReducedDim(myoblast_combined_sce,
dimred = "fastmnn_UMAP",
colour_by = "ENSG00000196090", #PTPRT
color_by = "ENSG00000196090", #PTPRT
point_size= 0.5,
point_alpha = 0.4,
other_fields = "diagnosis_group") +
Expand Down Expand Up @@ -720,7 +719,7 @@ scater::plotExpression(tumor_sce,
# a vector of genes to plot
features = genes_to_plot,
x = "diagnosis_group",
colour_by = "diagnosis_group",
color_by = "diagnosis_group",
other_fields = "celltype_broad",
point_size = 0.1) +
# each celltype is its own column
Expand All @@ -729,7 +728,7 @@ scater::plotExpression(tumor_sce,
rows = vars(Feature)) +
# change the font size of the facet labels
theme(strip.text = element_text(size = 7)) +
guides(colour = guide_legend(
guides(color = guide_legend(
title = "Subtype", # update the legend title
# change the size of the legend colors
override.aes = list(size = 3, alpha = 1))
Expand Down
4 changes: 1 addition & 3 deletions scRNA-seq-advanced/05-gene_set_enrichment_analysis.Rmd
View file
Original file line number Diff line number Diff line change
Expand Up @@ -78,9 +78,7 @@ hodgkins_analysis_dir <- file.path("analysis", "hodgkins")
# We'll create a directory to specifically hold the pathway results if it doesn't
# exist yet
results_dir <- file.path(hodgkins_analysis_dir, "pathway-analysis")
if (!dir.exists(results_dir)) {
dir.create(results_dir, recursive = TRUE)
}
fs::dir_create(results_dir)
```

#### Input files
Expand Down

0 comments on commit b5d6474

Please sign in to comment.