Hex code update from adding cancer_group as display_group

[kgaonkar6]

Purpose/implementation Section

What scientific question is your analysis addressing?

Add updated hex codes for plotting groups.

What was your approach?

Updated figures/mapping-histology-labels.Rmd to generate hex codes for cancer_group which was extracted from harmonized_diagnosis, additionally we removed benign and non-tumors and any group which has less than 5 counts

What GitHub issue does your pull request address?

#1140

Directions for reviewers. Tell potential reviewers what kind of feedback you are soliciting.

Which areas should receive a particularly close look?

NA

Is there anything that you want to discuss further?

NA

Is the analysis in a mature enough form that the resulting figure(s) and/or table(s) are ready for review?

Yes

Results

What types of results are included (e.g., table, figure)?

table

What is your summary of the results?

38 cancer_group are now available.

Reproducibility Checklist

• The dependencies required to run the code in this pull request have been added to the project Dockerfile.
• This analysis has been added to continuous integration.

Documentation Checklist

• This analysis module has a README and it is up to date.
• This analysis is recorded in the table in analyses/README.md and the entry is up to date.
• The analytical code is documented and contains comments.

[kgaonkar6]

CI fails because of "High-grade glioma/astrocytoma" used as the filename which is not valid since the "/" in unix denotes the sub-folder "astrocytoma" which does not exist. We will update to display_group in #1142

[jharenza]

Just want to be sure we are keeping display group as is and adding cancer group as a separate set of hex codes. Can check in later.

[kgaonkar6]

This is the description of the 2 columns in the output file:

OpenPBTA-analysis/figures/mapping-histology-labels.Rmd

Lines 28 to 29 in 13b5725

	- `display_group` - the high-level histology labels that should be used for plotting
	- `hex_codes` the direct colors that should be used for plotting

My understanding from the comment here was that we will use cancer_group as the disease label/grouping (ie. display_group right?) #917 (comment).

But did you want to use 1 set of hex_codes for display_group ( which in master branch which is extracted from broad_histology) and another for set of hex_codes for cancer_group ? Then we will update plotting scripts in oncoprint/mutsig to use the cancer_group column values instead of values in display_group column?

[jharenza]

I started reviewing some of the figures and I think we need to keep some at a broader display group, have some be annotated with cancer group, but may have some annotated with both, so hex codes for both, yes. Sorry for the confusion - I had not dove into the figures when making that comment, but started making notes in the GSlides doc here and making a ticket here #1144. For now, I would say simply add a new cancer_group with respective hex codes.

That being said, I know we were talking about not including colors for some benign/ N<5 groups because those groupings are too low for group analyses. However, when looking at the transcriptomic reduction figure, I think that we should not remove any samples from that, for example. I think we might have to do the filtering within modules, rather than in this hex code module - does that make sense?

[kgaonkar6]

I started reviewing some of the figures and I think we need to keep some at a broader display group, have some be annotated with cancer group, but may have some annotated with both, so hex codes for both, yes. Sorry for the confusion - I had not dove into the figures when making that comment, but started making notes in the GSlides doc here and making a ticket here #1144. For now, I would say simply add a new cancer_group with respective hex codes.

Thank you for the clarification. I believe my code update in c2a9392 satisfies the requirements above.

That being said, I know we were talking about not including colors for some benign/ N<5 groups because those groupings are too low for group analyses. However, when looking at the transcriptomic reduction figure, I think that we should not remove any samples from that, for example. I think we might have to do the filtering within modules, rather than in this hex code module - does that make sense?

That makes sense, I'll remove the <5 filter .

However cancer_group is NA for Other, Benign tumor, Dysplasia/Gliosis and Normal samples.
Should I update the hex color codes to be filtered for tumor samples , so when cancer_group==NA we know they are Other, Benign tumor, Dysplasia/Gliosis so we can add the gray hex code.
Note Normal samples will be removed from the hex_code assignment I don't see a use-case for the normal samples getting a cancer_group_hex_code.

Currently for display_group in master (extracted from broad_histologies) adds a gray color for benign and other tumors and black for Normal samples.

[jharenza]

Should I update the hex color codes to be filtered for tumor samples , so when cancer_group==NA we know they are Other, Benign tumor, Dysplasia/Gliosis so we can add the gray hex code.

This sounds good for consistency and agree with not having normals in there

[jharenza]

I am going to test out some other color palettes to try to make sure the new colors are cohesive with the display_group, but in the meantime, made some other notes.

[jharenza]

Suggested change

	# Keep only cancer_groups with 5 or more counts
	dplyr::filter(cancer_group_n >=5,
	# remove NA
	!is.na(cancer_group)) %>%
	# Remove NA
	dplyr::filter(!is.na(cancer_group)) %>%

[kgaonkar6]

We are not removing the cancer_group NA now but keeping them and adding a gray color

[jharenza]

per slack discussion, @kgaonkar6 will update cancer group hex codes to:

"#ff0000",
"#f20000",
"#997373",
"#403030",
"#330700",
"#ff9180",
"#591800",
"#b2502d",
"#cca799",
"#ff6600",
"#ffb380",
"#7f5940",
"#cc6d00",
"#331b00",
"#ccb499",
"#ffaa00",
"#996600",
"#594316",
"#ffd580",
"#ffee00",
"#998f00",
"#999673",
"#303300",
"#fbffbf",
"#ccff00",
"#494d39",
"#b5d96c",
"#6a8040",
"#66ff00",
"#42a600",
"#bfffbf",
"#003307",
"#00661b",
"#00ff88",
"#86b39e",
"#00b377",
"#006652",
"#00ffee",
"#00a7b3",
"#bffbff",
"#567173",
"#00ccff",
"#003d4d",
"#00aaff",
"#267399",
"#0088ff",
"#0042a6",
"#001a40",
"#bfd9ff",
"#0044ff",
"#394973",
"#000e66",
"#bfbfff",
"#9180ff",
"#5800a6",
"#754d99",
"#aa00ff",
"#3a3040",
"#aa86b3",
"#530059",
"#ff00ee",
"#a60085",
"#330022",
"#ff80d5",
"#ff0088",
"#804062",
"#a60042",
"#590024",
"#ffbfd9",
"#ff0044",
"#990014",
"#ff8091"

[kgaonkar6]

Thanks for the colors @jharenza ! I've updated the colors and I'm using v21 histologies file now.

[jharenza]

colors look better

[jaclyn-taroni]

Can I ask why this is marked with blocked and don't merge? Do we want #1157 to go in first?

[jharenza]

The cancer groups are being updated with the new histology file (we had GNG and Ganglioglioma separately and there were some issues with the / in the names.