Last updated: 2026-09-09

Checks: 7 0

Knit directory: immgenT-GP-analysis/analysis/

This reproducible R Markdown analysis was created with workflowr (version 1.7.2). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.

R Markdown file: up-to-date

Great! Since the R Markdown file has been committed to the Git repository, you know the exact version of the code that produced these results.

Environment: empty

Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.

Seed: set.seed(1)

The command set.seed(1) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.

Session information: recorded

Great job! Recording the operating system, R version, and package versions is critical for reproducibility.

Cache: none

Nice! There were no cached chunks for this analysis, so you can be confident that you successfully produced the results during this run.

File paths: relative

Great job! Using relative paths to the files within your workflowr project makes it easier to run your code on other machines.

Repository version: 9fab2f6

Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility.

The results in this page were generated with repository version 9fab2f6. See the Past versions tab to see a history of the changes made to the R Markdown and HTML files.

Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated: 


Ignored files:
    Ignored:    .DS_Store
    Ignored:    .claude/
    Ignored:    analysis/.DS_Store
    Ignored:    analysis/.Rhistory
    Ignored:    analysis/assets/.DS_Store
    Ignored:    analysis/site_libs/
    Ignored:    captions/
    Ignored:    code/.DS_Store
    Ignored:    code/other/topic_flashier_20250212.R
    Ignored:    code/other/topic_wrapper_20250215_alldata_backfit.sh
    Ignored:    data
    Ignored:    experiments/
    Ignored:    figures/.DS_Store
    Ignored:    figures/Previous/.DS_Store
    Ignored:    figures/Previous/bits/.DS_Store
    Ignored:    figures/Previous/bits/Figure 1/.DS_Store
    Ignored:    figures/Previous/bits/Figure 2/.DS_Store
    Ignored:    figures/Previous/bits/Figure 3/.DS_Store
    Ignored:    figures/Previous/bits/Figure 4/.DS_Store
    Ignored:    figures/Previous/bits/Figure 6/.DS_Store
    Ignored:    figures/Previous/bits/Figure 7/.DS_Store
    Ignored:    figures/Previous/bits/Figure S1/.DS_Store
    Ignored:    figures/Previous/bits/Figure S2/.DS_Store
    Ignored:    figures/Previous/bits/Figure S3/.DS_Store
    Ignored:    figures/Previous/bits/Figure S6/.DS_Store
    Ignored:    figures/Previous/bits/Figure S7/.DS_Store
    Ignored:    figures/final-selected/.DS_Store
    Ignored:    figures/final-selected/Figure 1/.DS_Store
    Ignored:    figures/final-selected/Figure 2/.DS_Store
    Ignored:    figures/final-selected/Figure 4/.DS_Store
    Ignored:    figures/final-selected/Figure S1/.DS_Store
    Ignored:    figures/templates_20260729/
    Ignored:    internal/
    Ignored:    log/
    Ignored:    output/.DS_Store
    Ignored:    output/Figure2/
    Ignored:    output/Figure7b/7b_cell_metadata.csv.gz
    Ignored:    plan/
    Ignored:    tables/
    Ignored:    tmp/

Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.

These are the previous versions of the repository in which changes were made to the R Markdown (analysis/FigureS5.Rmd) and HTML (docs/FigureS5.html) files. If you’ve configured a remote Git repository (see ?wflow_git_remote), click on the hyperlinks in the table below to view the files as they were in that past version.

File Version Author Date Message
Rmd 9fab2f6 Ziang Zhang 2026-09-09 Extended Data Figure 5: show the union of both tissue-associated GP sets
html b0c1d19 Ziang Zhang 2026-09-09 Build site: Extended Data Figure 5b recoloured
Rmd 5be33df Ziang Zhang 2026-09-09 Extended Data Figure 5b: purple is the positive end, green the negative
html a7a481f Ziang Zhang 2026-09-09 Build site: the rebuilt Figure 1d and the Extended Data renumbering
Rmd c233cd8 Ziang Zhang 2026-09-09 Extended Data reorganisation: split the tissue/cluster figure, renumber 3-8
html 1e88d7e Ziang Zhang 2026-09-04 Build site: published captions and titles across all 24 pages
Rmd 0267e5b Ziang Zhang 2026-09-04 Captions from the published manuscript; trim editor notes off the page code
html 5d3b86c Ziang Zhang 2026-09-03 Build site: Extended Data Figure 5 page rebuilt after the recolouring
Rmd 2101847 Ziang Zhang 2026-09-03 Extended Data Figure 5: colour each lineage row on its own
html 19c977f Ziang Zhang 2026-09-02 Build site: Extended Data 5-7 renumbered, Figure S5 page rebuilt
Rmd 1e5721a Ziang Zhang 2026-09-02 Extended Data 5-7 renumbered, and Figure S5 assembled as one stacked figure
html ae03072 Ziang Zhang 2026-08-19 Build site: six pages rebuilt after the prose cleanup
Rmd adc2327 Ziang Zhang 2026-08-19 Site prose: finish taking internal notes off the pages
html eeca07b Ziang Zhang 2026-08-05 Keep pre-refactor provenance in panel comments off the published pages
Rmd 5651d0e Ziang Zhang 2026-08-05 Extended Data tables: reorder to six, rebuild Table 1, drop internal notes
html cbcec52 Ziang Zhang 2026-07-30 Build site: Extended Data Figure naming
Rmd 66aa029 Ziang Zhang 2026-07-30 Name the Extended Data figures as published on the site
html ac650a0 Ziang Zhang 2026-07-30 Build site: Figure S5 (ex-S6a) and Figure S6 as a-f
Rmd c9b020f Ziang Zhang 2026-07-30 Split Figure S6’s protein-program heatmap out as Figure S5
html d538aa2 Ziang Zhang 2026-07-28 Build site: reordered Figures 6 / S6 / S3 and the new Figure 7b page
Rmd 4c07670 Ziang Zhang 2026-07-28 Reorder Figures 6, S6 and S3; make the ex-S5 figure Figure 7b
html 029b0ae Ziang Zhang 2026-07-28 Build site.
Rmd 0f5b5da Ziang Zhang 2026-07-28 Align all figure captions with captions_20260728_final.docx
html 3fc3789 Ziang Zhang 2026-07-27 Republish all 24 pages
html 1390a03 Ziang Zhang 2026-07-27 Republish all 24 pages
html adaef21 Ziang Zhang 2026-07-27 Build site: panel fixes and PDF-derived assets
html 5b19858 Ziang Zhang 2026-07-27 Build site.
Rmd ffe285c Ziang Zhang 2026-07-27 Reorganize figures/ and untrack local-only exploration notes
html f7d90e7 Ziang Zhang 2026-07-26 Build site.
Rmd b0b2b2f Ziang Zhang 2026-07-26 Tidy Figure S5 page: name the S5a/S5b/colorbar panels
html fe93d0d Ziang Zhang 2026-07-23 Build site.
Rmd 61de7cb Ziang Zhang 2026-07-23 Reflect single-matching pipeline on the Figure S5 page
html 9862b6d Ziang Zhang 2026-07-23 Build site.
Rmd 98d2924 Ziang Zhang 2026-07-23 Reword Figure S5 page for a publication audience
html 7ddbdb4 Ziang Zhang 2026-07-23 Build site.
Rmd b138063 Ziang Zhang 2026-07-23 Reformat Figure S5 page: lead with the figure, concise methods, link
html 9398c72 Ziang Zhang 2026-07-23 Publish Figure S5 workflowr page
Rmd b9f4f58 Ziang Zhang 2026-07-23 Add Figure S5: EBMF vs matched-RQVI level2-cluster comparison

Both panels are produced by script/FigureS5.R. They show the same 32 GPs – every program the manuscript calls tissue-associated – first across the 18 tissues and then across the 8 T cell lineages, so that a program’s tissue distribution can be read against its lineage distribution on the same rows. The analysis uses healthy non-thymocyte cells (condition_broad == "healthy" and annotation_level1 != "thymocyte"). For each GP, its mean loading across the groups shown is subtracted from every group mean, so a row says where a program is more or less active than its own average rather than how large its loading is. It is the tissue counterpart of Extended Data Figure 3, which shows all 200 GPs across the level-2 clusters, and it puts the tissue-associated programs behind Figure 5’s AUC analysis on a single pair of axes.

Setup and the 32 tissue-associated GPs

The rows are the union of the two criteria by which a GP is called tissue-associated:

  • activity – its raw (uncentered) mean loading reaches 0.1 in at least one of the 18 tissues. 31 of the 200 GPs pass. This is a threshold on average activity, so a program carried by a small number of cells fails it.
  • discrimination – its one-vs-rest tissue AUC exceeds 0.9, under the same positivity mask Figure 5 uses (mean loading in the tissue above the GP’s overall mean) and with tissues under 100 cells dropped. Seven GPs pass; these are the seven Figure 5 highlights throughout.

Six GPs satisfy both, so 32 are shown. GP37 is the only one selected by discrimination alone: its mammary-gland AUC is 0.9918, but it is active in 2.5% of mammary gland cells, which dilutes its mean loading there to 0.011 – so its row is near-white in both panels by construction. The AUC criterion is re-derived here and checked against the GP list in script/Figure5.R, and the union is checked against a literal list, so a change in the loadings, the AUCs, the cell filter or either cutoff fails the run rather than quietly redrawing a different figure:

# Figure S5. The tissue-associated GPs, across tissues and across lineages.
#
# Panels produced (see analysis/FigureS5.Rmd for the caption text):
#   s5a  Row-centered mean GP activity across the 18 tissues, restricted to the
#        32 tissue-associated GPs. Blue-white-red scale.
#   s5b  The same 32 GPs in the same row order, but across the 8 T cell
#        lineages. Green-white-purple scale (purple positive), so the two halves
#        of the figure cannot be mistaken for each other.
#
#
# The rows are the union of the two tissue-associated GP sets the manuscript
# names -- see "GP selection" below. The panels then show, per GP, how that
# activity is distributed, by subtracting the GP's mean across the groups
# shown from every group mean. Each panel's centered color scale is fixed:
# [-0.2, 0.2] for (a), matching the retired 200-GP tissue heatmap it comes from,
# and the tighter [-0.1, 0.1] for (b), because spreading a program over eight
# lineages instead of eighteen tissues gives much smaller deviations and (a)'s
# scale renders the lineage panel almost blank. Values outside each range
# saturate at the endpoint colors: 0.9% of (a)'s cells and 3.6% of (b)'s.
#
# Required inputs (data/) -- see code/README.md's "Data provenance" table
# for the full picture:
#   L_pm_filtered.rds                        [code/pipeline/01b_filter_cells.R]
#   igt1_96_..._ADTonly.Rds                  [primary input Seurat object]
#   organ_simplified_AUC_list_figure_no_thymocytes_healthy.rds
#                                            [code/pipeline/02_compute_auc.R]

suppressPackageStartupMessages({
  library(ComplexHeatmap)
  library(circlize)
  library(grid)
  library(ZemmourLib)
})

if (!file.exists("code/R/setup_data.R")) {
  stop("Run this script from the immgenT-GP-analysis repository root.")
}

source("code/R/setup_data.R")
source("code/R/centered_mean_heatmap.R")

figure_path <- "figures/final-selected/Figure S5"
dir.create(figure_path, recursive = TRUE, showWarnings = FALSE)
run_started_at <- Sys.time()

# ============================================================
# Setup: the healthy non-thymocyte tissue and lineage mean matrices
# ============================================================
gp_data <- load_gp_data()
reference <- healthy_nonthymocyte_reference(gp_data)
L_reference <- reference$L
meta_reference <- reference$meta

organ_color_limit <- 0.2
level1_color_limit <- 0.1
level1_order <- c("CD8", "CD4", "Treg", "gdT", "CD8aa", "Tz", "DN", "DP")

organ_raw <- mean_loading_by_group(L_reference, meta_reference$organ_simplified)$matrix
level1_raw <- mean_loading_by_group(L_reference, meta_reference$annotation_level1)$matrix

# ============================================================
# GP selection: the union of the two tissue-associated GP sets
# ============================================================
# The manuscript names two, by different criteria, and this figure shows both:
#
#   A  "31 programs active across tissues" -- raw (uncentered) mean loading
#      reaches 0.1 in at least one of the 18 tissues. The filter from
#      experiments/healthy_nonthymus_mean_loading_heatmaps/. It is a threshold
#      on *average* activity, so a program carried by a few cells fails it.
#   B  "Seven GPs were strongly tissue-associated (AUC > 0.9)" -- Figure 5's
#      rule: one-vs-rest tissue AUC above 0.9 under Figure 5's positivity mask
#      (mean loading in the tissue above the GP's overall mean), with tissues
#      under 100 cells dropped.
#
# Six GPs satisfy both, so the union is 32. GP37 is the only member of B alone:
# its mammary-gland AUC is 0.9918, but it is active in 2.5% of mammary gland
# cells, which dilutes its mean loading there to 0.011 -- an order of magnitude
# under rule A's cutoff. Its row is therefore near-white in (a) by construction.
#
# Rule B is re-derived here rather than retyped, and checked against the literal
# `gps_of_interest` in script/Figure5.R, so the two figures cannot disagree
# about which GPs are the strongly tissue-associated ones. The union is checked
# against a literal list as well, so a change in the loadings, the AUCs, the
# cell filter or either cutoff fails here instead of quietly redrawing a
# different figure.
raw_mean_cutoff <- 0.1
auc_cutoff <- 0.9
min_tissue_cells <- 100L

# --- A: mean activity ---
tissue_active <- rowSums(organ_raw >= raw_mean_cutoff) > 0L

# --- B: tissue discrimination, on Figure 5's terms ---
organ_auc <- readRDS(paste0(
  "data/", "organ_simplified_AUC_list_figure_no_thymocytes_healthy.rds"
))$auc
colnames(organ_auc) <- paste0("GP", seq_len(ncol(organ_auc)))
tissue_counts <- table(meta_reference$organ_simplified)
large_tissues <- setdiff(
  rownames(organ_auc), names(tissue_counts[tissue_counts < min_tissue_cells])
)
organ_auc <- organ_auc[large_tissues, , drop = FALSE]

overall_mean <- colMeans(L_reference, na.rm = TRUE)
tissue_mean <- t(vapply(
  large_tissues,
  function(tissue) {
    colMeans(L_reference[meta_reference$organ_simplified == tissue, , drop = FALSE], na.rm = TRUE)
  },
  numeric(ncol(L_reference))
))
positive_mask <- sweep(tissue_mean, 2L, overall_mean, "-") > 0
tissue_specific <- colSums((organ_auc > auc_cutoff) & positive_mask) > 0L
tissue_specific <- rownames(organ_raw) %in% colnames(organ_auc)[tissue_specific]

# Figure 5 highlights these seven throughout; rule B must reproduce exactly them.
figure5_gps_of_interest <- local({
  lines <- readLines("script/Figure5.R", warn = FALSE)
  hit <- grep("^gps_of_interest <- ", lines, value = TRUE)
  if (length(hit) != 1L) {
    stop("gps_of_interest is not assigned exactly once in script/Figure5.R")
  }
  eval(parse(text = sub("^gps_of_interest <- ", "", hit)))
})
stopifnot(setequal(rownames(organ_raw)[tissue_specific], figure5_gps_of_interest))

selected <- tissue_active | tissue_specific
selected_gps <- rownames(organ_raw)[selected]

expected_gps <- paste0("GP", c(
  1, 3, 4, 6, 8, 9, 11, 22, 23, 25, 26, 29, 30, 32, 35, 37, 41, 43, 49, 51, 58,
  62, 63, 72, 80, 93, 100, 166, 170, 171, 174, 177
))
stopifnot(
  sum(tissue_active) == 31L,
  sum(tissue_specific) == 7L,
  length(expected_gps) == 32L,
  identical(selected_gps, expected_gps)
)

organ_selected_raw <- organ_raw[selected, , drop = FALSE]
level1_selected_raw <- level1_raw[selected, , drop = FALSE]
organ_centered <- center_by_gp_mean(organ_selected_raw)
level1_centered <- center_by_gp_mean(level1_selected_raw)

# ============================================================
# Ordering: (a) sets the row order, (b) reuses it
# ============================================================
# Recomputing the dominant-group order on the 32-row submatrix (rather than
# inheriting the retired 200-GP order) is what makes the tissue columns reflect
# the GPs actually on display. Panel (b) then keeps (a)'s row order, so a GP sits
# on the same line in both halves and can be read across; its columns are
# Figure 1's lineage order rather than a dominant-group order.
organ_order <- dominant_group_order(organ_selected_raw)

level1_groups <- colnames(level1_centered)
if (!setequal(level1_groups, level1_order)) {
  stop("The observed lineages do not match the Figure 1 level1 order.")
}
level1_column_order <- order(match(level1_groups, level1_order))
level1_row_order <- match(
  rownames(organ_centered)[organ_order$row_order],
  rownames(level1_centered)
)

stopifnot(
  nrow(organ_centered) == 32L,
  nrow(level1_centered) == 32L,
  ncol(organ_centered) == 18L,
  ncol(level1_centered) == 8L,
  identical(rownames(organ_centered), rownames(level1_centered)),
  max(abs(rowMeans(organ_centered))) < 1e-12,
  max(abs(rowMeans(level1_centered))) < 1e-12
)

organ_palette <- palette_for_groups(
  colnames(organ_centered),
  ZemmourLib::immgent_colors$organ_simplified,
  "organ_simplified"
)
level1_palette <- palette_for_groups(
  colnames(level1_centered),
  ZemmourLib::immgent_colors$level1,
  "annotation_level1"
)

(a) Tissue mean loading

# ============================================================
# s5a: the 31 tissue-active GPs across the 18 tissues
# ============================================================
render_centered_heatmap(
  organ_centered,
  organ_palette,
  "tissue (organ_simplified)",
  file.path(figure_path, "s5a.pdf"),
  organ_order$row_order,
  organ_order$column_order,
  organ_color_limit,
  "32 tissue-associated GPs; dominant-group blocks",
  palette = heatmap_palettes$blue_red
)

Version Author Date
9fab2f6 Ziang Zhang 2026-09-09
c233cd8 Ziang Zhang 2026-09-09

Extended Data Fig. 5a. Row-centered mean GP activity across 18 tissues, for the 32 tissue-associated GPs. Mean GP activity computed in samples at baseline.

(b) Lineage mean loading

# ============================================================
# s5b: the same 31 GPs across the 8 lineages
# ============================================================
render_centered_heatmap(
  level1_centered,
  level1_palette,
  "lineage (annotation_level1)",
  file.path(figure_path, "s5b.pdf"),
  level1_row_order,
  level1_column_order,
  level1_color_limit,
  "the same 32 GPs, row order from (a)",
  palette = heatmap_palettes$green_purple
)

Version Author Date
9fab2f6 Ziang Zhang 2026-09-09
5be33df Ziang Zhang 2026-09-09
c233cd8 Ziang Zhang 2026-09-09

Extended Data Fig. 5b. The same 32 GPs across the 8 T cell lineages, in the row order of (a). Mean GP activity computed in samples at baseline. Note the different color scale: green-purple rather than blue-red, and -0.1 to 0.1 rather than -0.2 to 0.2, since spreading a program over 8 lineages instead of 18 tissues gives correspondingly smaller deviations.

What the panels drew

The 32 GPs, which criterion selected each, their dominant tissue and lineage, and their row position are written out alongside the figure:

# ============================================================
# What the panels drew
# ============================================================
record_dir <- "output/FigureS5"   # build intermediate (not a manuscript panel)
dir.create(record_dir, recursive = TRUE, showWarnings = FALSE)

write.csv(
  data.frame(
    GP = rownames(organ_centered),
    max_raw_tissue_mean = as.numeric(apply(organ_selected_raw, 1L, max)),
    dominant_tissue = colnames(organ_selected_raw)[max.col(organ_selected_raw, ties.method = "first")],
    dominant_lineage = colnames(level1_selected_raw)[max.col(level1_selected_raw, ties.method = "first")],
    max_masked_tissue_auc = as.numeric(apply(
      ifelse(positive_mask[, selected, drop = FALSE], organ_auc[, selected, drop = FALSE], NA_real_),
      2L, max, na.rm = TRUE
    )),
    rule = ifelse(
      tissue_active[selected] & tissue_specific[selected], "A+B",
      ifelse(tissue_active[selected], "A", "B")
    ),
    row_in_panel = match(rownames(organ_centered), rownames(organ_centered)[organ_order$row_order])
  ),
  file.path(record_dir, "s5_selected_gps.csv"),
  row.names = FALSE,
  quote = FALSE
)

write.csv(
  data.frame(
    panel = c("s5a", "s5b"),
    grouping = c("organ_simplified", "annotation_level1"),
    view = "row-centered mean loading of the 32 tissue-associated GPs",
    gp_count = c(nrow(organ_centered), nrow(level1_centered)),
    group_count = c(ncol(organ_centered), ncol(level1_centered)),
    selection = paste0("raw mean loading >= ", raw_mean_cutoff,
                       " in >= 1 tissue OR masked tissue AUC > ", auc_cutoff),
    centered_definition = "group mean minus mean across groups for each GP",
    palette = c("blue-white-red", "green-white-purple"),
    color_min = c(-organ_color_limit, -level1_color_limit),
    color_mid = 0,
    color_max = c(organ_color_limit, level1_color_limit),
    observed_min = c(min(organ_centered), min(level1_centered)),
    observed_max = c(max(organ_centered), max(level1_centered)),
    frac_saturated = c(mean(abs(organ_centered) > organ_color_limit),
                       mean(abs(level1_centered) > level1_color_limit))
  ),
  file.path(record_dir, "S5_summary.csv"),
  row.names = FALSE,
  quote = FALSE
)

# A panel that fails to write leaves the previous PDF in place and the script
# still exits 0, so check the files rather than the exit code.
expected_panels <- file.path(figure_path, c("s5a.pdf", "s5b.pdf"))
stale <- expected_panels[!file.exists(expected_panels) |
                           file.mtime(expected_panels) < run_started_at |
                           file.size(expected_panels) == 0]
if (length(stale) > 0L) {
  stop("These panels were not written by this run: ", paste(stale, collapse = ", "))
}

message("Wrote Figure S5 to ", normalizePath(figure_path))

sessionInfo()
R version 4.5.1 (2025-06-13)
Platform: aarch64-apple-darwin20
Running under: macOS Sequoia 15.6.1

Matrix products: default
BLAS:   /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib 
LAPACK: /Library/Frameworks/R.framework/Versions/4.5-arm64/Resources/lib/libRlapack.dylib;  LAPACK version 3.12.1

locale:
[1] en_CA/en_CA/en_CA/C/en_CA/en_CA

time zone: America/Chicago
tzcode source: internal

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

loaded via a namespace (and not attached):
 [1] vctrs_0.7.3     cli_3.6.6       knitr_1.50      rlang_1.2.0    
 [5] xfun_0.55       stringi_1.8.9   otel_0.2.0      promises_1.5.0 
 [9] jsonlite_2.0.0  workflowr_1.7.2 glue_1.8.1      rprojroot_2.1.1
[13] git2r_0.36.2    htmltools_0.5.9 httpuv_1.6.16   sass_0.4.10    
[17] rmarkdown_2.30  evaluate_1.0.5  jquerylib_0.1.4 tibble_3.3.0   
[21] fastmap_1.2.0   yaml_2.3.12     lifecycle_1.0.5 whisker_0.4.1  
[25] stringr_1.6.0   compiler_4.5.1  fs_1.6.6        Rcpp_1.1.1-1.1 
[29] pkgconfig_2.0.3 later_1.4.4     digest_0.6.39   R6_2.6.1       
[33] pillar_1.11.1   magrittr_2.0.5  bslib_0.9.0     tools_4.5.1    
[37] cachem_1.1.0