Complete draft of first spatial notebook by sjspielman · Pull Request #949 · AlexsLemonade/training-modules

sjspielman · 2026-03-03T17:44:20Z

Closes #923
Closes #924

This PR adds the remaining sections to the first spatial notebook draft. I aimed to keep these sections pretty short since we have a lot already!

Section on normalization: This is very brief, and pretty straightforward. Anything else we'd like to see here?
Section on pretty plots: For this section...
- I defined a vector of genes of interest, alongside a table briefly describing those genes. I then plot expression for two of the genes. Note that I'm using indexing here with the goi vector I defined - let me know if that's too confusing and I should hardcode some more.
- Then, I have some TBD: either go on your own adventure to pick a gene from our list and plot it, or we can show them how to plot it all at once with purrr. I like the independent exploration on one hand, but I also really appreciate the story we can tell from seeing all these plots at once. Anyone have a preference?
- Note: I only included plotCoords() here, since plotVisium() with overlaid colored spots is a wild adventure to interpret. Does anyone want to advocate for showing it (even if only to show "gee I can't read this")?
- Cool with regular viridis? Of course this is logcounts and not a z-score, but I'm not worried about reusing that palette in this case. Does anyone disagree?

Rendered notebook:
knitted.nb.html

jashapiro

Overall this looks good. I mostly have some comments about the gene list and how to handle getting it. It would be nice to provide the code to look up the gene ids from gene names (even better if that is already in the Space Ranger-derived rowData, but I don't remember if it is right now). But I would only use a vector after plotting a couple of individual genes (not pulling them from the vector). Or if you do use a named vector, use the names!

jashapiro · 2026-03-04T23:43:39Z

spatial/01-spatial_intro.Rmd

+goi_list <- c(
+  "WT1" = "ENSG00000184937",
+  "SIX1" = "ENSG00000126778",
+  "MYCN" = "ENSG00000134323",
+  "NCAM1" = "ENSG00000149294",
+  "COL1A1" = "ENSG00000108821",
+  "TAGLN" = "ENSG00000149591", 
+  "PECAM1" = "ENSG00000261371"
+)


I do think this is overkill. Just show the two genes. If you want to get fancy, maybe show them how to look up the gene symbol for a gene of interest, rather than providing a named vector (which is not a list, so don't name it that!)

jashapiro · 2026-03-04T23:48:22Z

spatial/01-spatial_intro.Rmd

+We have a few more genes listed in the table, take a moment to plot one (or more!) below.
+Do you see any spatial structure that seems consistent with the H&E? 
+Different from the H&E?
+Are there regions of shared expression among genes, and would you expect that based on WT biology?


I won't go quietly.

Suggested change

Are there regions of shared expression among genes, and would you expect that based on WT biology?

Are there regions of shared expression among genes, and would you expect that based on Wilms tumor biology?

jashapiro · 2026-03-04T23:49:04Z

spatial/01-spatial_intro.Rmd

+
+**OR**
+
+Let's use some `purrr` magic to plot this all at once!


On the other hand, I'm not too mad at this. But I'd still want to start with just a list of gene names and get the ensembl ids programmatically rather than providing them together.

jashapiro · 2026-03-04T23:56:01Z

spatial/01-spatial_intro.Rmd

+gene_index <- 1
+
+wt1_plot <- ggspavis::plotCoords(
+  norm_spe, 
+  # name of variable to color by
+  annotate = goi_list[gene_index],
+  # counts are plotted by default, so we need to override this
+  assay = "logcounts", 
+  point_size = 1
+) + 
+  ggplot2::scale_color_viridis_c() +
+  ggplot2::labs(title = names(goi_list[gene_index]))


If you do decide you want to use the vector of genes, don't make a named vector and then not take advantage of it!

Suggested change

gene_index <- 1

wt1_plot <- ggspavis::plotCoords(

norm_spe,

# name of variable to color by

annotate = goi_list[gene_index],

# counts are plotted by default, so we need to override this

assay = "logcounts",

point_size = 1

) +

ggplot2::scale_color_viridis_c() +

ggplot2::labs(title = names(goi_list[gene_index]))

gene_name <- "WT1"

wt1_plot <- ggspavis::plotCoords(

norm_spe,

# name of variable to color by

annotate = goi_list[gene_name],

# counts are plotted by default, so we need to override this

assay = "logcounts",

point_size = 1

) +

ggplot2::scale_color_viridis_c() +

ggplot2::labs(title = gene_name)

🤦‍♀️

sjspielman · 2026-03-05T14:30:06Z

I think what I would like to do is:

show plotting two genes individually (sans vector) that show contrasting biology. focus here on plotting, dont introduce potential confusion from a vector yet
then after we've plotting some, programmatically define a bigger vector of genes and show plotting them all with map, which demos a code strategies for looking at a lot of genes at once

allyhawkins

Just noting that I don't have any additional comments at this time beyond what's been posted and agree with the plan to look at two genes individually and then plot the vector of genes.

sjspielman · 2026-03-05T17:14:32Z

Updated as discussed! Current version: 01-spatial_intro.nb.html

allyhawkins

This looks good to me. My only comment is that I think the section showing the rowData and the two ensembl gene Ids is a little overkill and I would probably just remove it.

spatial/01-spatial_intro.Rmd

allyhawkins · 2026-03-05T18:32:42Z

spatial/01-spatial_intro.Rmd

+
+We'll need to provide the gene's corresponding row name, which is its Ensembl gene id, to `ggpsavis::plotCoords()` so let's go ahead and find those first.
+
+The `rowData` slot helpfully provides a gene symbol/Ensembl id mapping, if you haven't memorized all the ids yet:


allyhawkins · 2026-03-05T18:34:37Z

spatial/01-spatial_intro.Rmd

+
+
+```{r get-ensembl, live = TRUE}
+# define gene symbols


I agree that it can be helpful to show examples of grabbing information from the SPE, but I don't know that it's totally necessary here. Given the length of this notebook, I might favor skipping this and the next chunk.

Maybe put something like this in the exercise instead to demo for them? (with solution = TRUE)
I have a panel of plots in the exercise outline too so this could be ready to go.

sjspielman added 4 commits March 3, 2026 11:37

add normalization section

997b8eb

plot section

a086e7b

wording updates

3371af3

spelling

2216bd0

sjspielman marked this pull request as ready for review March 3, 2026 20:29

sjspielman requested review from allyhawkins and jashapiro March 3, 2026 20:29

jashapiro reviewed Mar 4, 2026

View reviewed changes

allyhawkins reviewed Mar 5, 2026

View reviewed changes

plot two genes standalone and then map

8e2228d

sjspielman requested review from allyhawkins and jashapiro March 5, 2026 17:14

allyhawkins approved these changes Mar 5, 2026

View reviewed changes

sjspielman added 2 commits March 9, 2026 17:35

fix bad link formatting and remove printing size factors

02072b5

combine chunks

ac9277b

	Are there regions of shared expression among genes, and would you expect that based on WT biology?
	Are there regions of shared expression among genes, and would you expect that based on Wilms tumor biology?


		OR

		Let's use some `purrr` magic to plot this all at once!


		We'll need to provide the gene's corresponding row name, which is its Ensembl gene id, to `ggpsavis::plotCoords()` so let's go ahead and find those first.

		The `rowData` slot helpfully provides a gene symbol/Ensembl id mapping, if you haven't memorized all the ids yet:

Uh oh!

Conversation

sjspielman commented Mar 3, 2026

Uh oh!

jashapiro left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sjspielman commented Mar 5, 2026

Uh oh!

allyhawkins left a comment

Choose a reason for hiding this comment

Uh oh!

sjspielman commented Mar 5, 2026

Uh oh!

allyhawkins left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants