Last updated: 2026-07-05
Checks: 6 1
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.
The R Markdown is untracked by Git. To know which version of the R
Markdown file created these results, you’ll want to first commit it to
the Git repo. If you’re still working on the analysis, you can ignore
this warning. When you’re finished, you can run
wflow_publish to commit the R Markdown file and build the
HTML.
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.
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.
Great job! Recording the operating system, R version, and package versions is critical for reproducibility.
Nice! There were no cached chunks for this analysis, so you can be confident that you successfully produced the results during this run.
Great job! Using relative paths to the files within your workflowr project makes it easier to run your code on other machines.
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 6d1c6c0. 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: code/.DS_Store
Ignored: data
Ignored: figures/final-selected/.DS_Store
Ignored: figures/final-selected/bits/.DS_Store
Ignored: figures/final-selected/bits/Figure 1/.DS_Store
Ignored: figures/final-selected/bits/Figure 2/.DS_Store
Ignored: figures/final-selected/bits/Figure 3/.DS_Store
Ignored: figures/final-selected/bits/Figure 4/.DS_Store
Ignored: figures/final-selected/bits/Figure 6/.DS_Store
Ignored: figures/final-selected/bits/Figure 7/.DS_Store
Ignored: figures/final-selected/bits/Figure S1/.DS_Store
Ignored: figures/final-selected/bits/Figure S2/.DS_Store
Ignored: figures/final-selected/bits/Figure S3/.DS_Store
Ignored: figures/final-selected/bits/Figure S6/.DS_Store
Ignored: figures/final-selected/bits/Figure S7/.DS_Store
Ignored: figures/generated/.DS_Store
Ignored: figures/generated/Figure 5/
Ignored: figures/generated/Figure 7/
Ignored: figures/generated/Figure 8/
Ignored: figures/generated/Figure 9/
Untracked files:
Untracked: analysis/ExtendedDataTable1.Rmd
Untracked: analysis/ExtendedDataTable2.Rmd
Untracked: analysis/ExtendedDataTable3.Rmd
Untracked: analysis/ExtendedDataTable4.Rmd
Untracked: analysis/ExtendedDataTable5.Rmd
Untracked: analysis/ExtendedDataTable6.Rmd
Untracked: analysis/ExtendedDataTable7.Rmd
Untracked: figures/generated/ExtendedDataTable1_GP_summary.csv
Untracked: figures/generated/ExtendedDataTable2_GP_AUC_lineage.csv
Untracked: figures/generated/ExtendedDataTable3_GP_AUC_tissue.csv
Untracked: figures/generated/ExtendedDataTable4_GP_AUC_cluster.csv
Untracked: figures/generated/ExtendedDataTable5_GP_during_activation.csv
Untracked: figures/generated/ExtendedDataTable6_protein_factor_matrix.csv
Untracked: figures/generated/ExtendedDataTable7_protein_gating.csv
Untracked: script/ExtendedDataTable1_GP_summary.R
Untracked: script/ExtendedDataTable2_GP_AUC_lineage.R
Untracked: script/ExtendedDataTable3_GP_AUC_tissue.R
Untracked: script/ExtendedDataTable4_GP_AUC_cluster.R
Untracked: script/ExtendedDataTable5_GP_during_activation.R
Untracked: script/ExtendedDataTable6_protein_factor_matrix.R
Untracked: script/ExtendedDataTable7_protein_gating.R
Untracked: tables/
Unstaged changes:
Deleted: analysis/TableS1.Rmd
Modified: analysis/index.Rmd
Modified: code/R/gated_protein_helpers.R
Modified: code/R/roc_auc.R
Modified: code/README.md
Modified: code/pipeline/02_compute_auc.R
Deleted: figures/generated/Supplementary_Table1_GP_summary.csv
Modified: script/README.md
Deleted: script/TableS1.R
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.
There are no past versions. Publish this analysis with
wflow_publish() to start tracking its development.
Produced by script/ExtendedDataTable1_GP_summary.R:
# Extended Data Table 1: summary of GP characteristics and annotations.
#
# One row per GP (GP1..GP200) with:
# - Lineage / Cluster / Tissue: the categories each GP *positively* predicts
# well -- one-vs-rest AUC > 0.8 AND the optimal decision threshold at or
# above the GP's median loading, so high loading (not low) drives the
# prediction. Categories are annotation_level1 (lineage), annotation_level2
# (cluster) and organ_simplified (tissue), all from the healthy
# non-thymocyte AUC family (*_no_thymocytes_healthy, same population as
# Figure 2/Figure 4); the median loading is likewise computed on the healthy
# non-thymocyte cells so the threshold comparison is on the same population.
# - Signature genes / proteins: the top 5 up- and top 5 down-regulated genes
# and proteins by factor score, among those with |score| > 0.1 on the
# max|.|=1-per-GP-scaled gene (F_pm_filtered) and protein (Figure 6's
# Protein_F_pm) factor matrices.
#
# Reworks the retired Table S1 (script/TableS1.R): drops its Condition column,
# switches the annotations from the non-thymocyte healthy+diseased AUC to the
# healthy non-thymocyte AUC, adds protein signatures, and loosens the gene
# signature cutoff from 0.25 to 0.1.
data_path <- "data/"
output_path <- "figures/generated/"
# ---- Protein factor matrix (Figure 6's Protein_F_pm: filtered + max|.|=1 per
# GP column). Load, extract, and drop the heavy fit before loading L. ----
Protein_flash_result <- readRDS(paste0(data_path, "protein_flash_selected_summary_lognorm_backfit200.rds"))
Protein_F_pm_raw <- Protein_flash_result$F_pm
rm(Protein_flash_result); gc()
isotype_proteins <- grep("^Isotype", rownames(Protein_F_pm_raw), value = TRUE)
proteins_quality <- read.csv(paste0(data_path, "TableS4_citeseq_qc_20250513.csv"), header = TRUE, stringsAsFactors = FALSE, skip = 1)
good_proteins <- c(proteins_quality$protein[proteins_quality$classification == "good"], "IL2RA.CD25", "ITB7", "CD69")
exclude_proteins <- c("CD19", "CD34", "CD45.1", "CD45.2", "CD138", "TCRVA2", "TER119")
thy11_proteins <- grep("THY1.1", rownames(Protein_F_pm_raw), value = TRUE)
Protein_F_pm <- Protein_F_pm_raw[!rownames(Protein_F_pm_raw) %in% isotype_proteins, ]
Protein_F_pm <- Protein_F_pm[rownames(Protein_F_pm) %in% good_proteins, ]
Protein_F_pm <- Protein_F_pm[!rownames(Protein_F_pm) %in% exclude_proteins, ]
Protein_F_pm <- Protein_F_pm[!rownames(Protein_F_pm) %in% thy11_proteins, ]
D_lognorm <- diag(1 / apply(Protein_F_pm, 2, function(x) max(abs(x))))
Protein_F_pm <- Protein_F_pm %*% D_lognorm
Protein_F_pm[is.na(Protein_F_pm)] <- 0
# ---- Loadings, gene factor matrix, healthy non-thymocyte AUC ----
# L_pm_filtered / F_pm_filtered and the cached AUC matrices all share the raw
# "K1","K2",... column names (from the flashier fit); keep them for matching and
# relabel to "GP1","GP2",... only for output.
L_pm_filtered <- readRDS(paste0(data_path, "L_pm_filtered.rds"))
F_pm_filtered <- readRDS(paste0(data_path, "F_pm_filtered.rds"))
level_1_AUC_list <- readRDS(paste0(data_path, "level_1_AUC_list_figure_no_thymocytes_healthy.rds"))
level_2_AUC_list <- readRDS(paste0(data_path, "level_2_AUC_list_figure_no_thymocytes_healthy.rds"))
organ_AUC_list <- readRDS(paste0(data_path, "organ_simplified_AUC_list_figure_no_thymocytes_healthy.rds"))
# Use L_pm_filtered's "K1".."K200" names as the canonical GP key: the AUC
# matrices carry these same names, whereas F_pm_filtered's raw columns are
# "F1".."F200" and the protein factor columns are unnamed -- but all three are
# in the same factor order as L (same flashier fit / fixed-loading projection),
# so relabeling them to L's K-names aligns every signature/annotation lookup.
gps <- colnames(L_pm_filtered) # "K1".."K200"
gp_labels <- paste0("GP", seq_along(gps))
colnames(Protein_F_pm) <- gps
# Normalize the gene factor matrix so each GP column has max|score| = 1.
F_pm_filtered <- apply(F_pm_filtered, 2, function(x) x / max(abs(x)))
colnames(F_pm_filtered) <- gps
# ---- Median loading on the healthy non-thymocyte cell subset ----
seurat_meta <- readRDS(paste0(data_path, "igt1_96_withtotalvi20260206_clean_ADTonly.Rds"))@meta.data
seurat_meta_filtered <- seurat_meta[rownames(L_pm_filtered), ]
rm(seurat_meta); gc()
# which() drops any cells whose condition/lineage metadata is NA (so they don't
# leak in as NA rows and turn every median into NA). seurat_meta_filtered is row-
# aligned to L_pm_filtered, so the integer index selects the matching L rows.
healthy_nonthy_idx <- which(
seurat_meta_filtered$condition_broad == "healthy" &
seurat_meta_filtered$annotation_level1 != "thymocyte"
)
L_healthy_nonthy <- L_pm_filtered[healthy_nonthy_idx, , drop = FALSE]
# ---- Positively-predicted categories: AUC > 0.8 and threshold >= median
# loading (so high loading drives the prediction) ----
get_passing_categories <- function(auc_list, gps, L_mat, threshold = 0.8) {
auc_mat <- auc_list$auc
thr_mat <- auc_list$threshold
gp_median_loading <- apply(L_mat, 2, median)
lapply(gps, function(gp) {
if (!gp %in% colnames(auc_mat)) return("")
auc_vals <- auc_mat[, gp]
thr_vals <- thr_mat[, gp]
med_load <- gp_median_loading[gp]
pass <- !is.na(auc_vals) & !is.na(thr_vals) &
auc_vals > threshold & thr_vals >= med_load
cats <- rownames(auc_mat)[pass]
paste(cats, collapse = "; ")
})
}
lineage_cats <- get_passing_categories(level_1_AUC_list, gps, L_healthy_nonthy)
cluster_cats <- get_passing_categories(level_2_AUC_list, gps, L_healthy_nonthy)
tissue_cats <- get_passing_categories(organ_AUC_list, gps, L_healthy_nonthy)
# ---- Top 5 up / down signature genes and proteins (|score| > 0.1) ----
top_signatures <- function(score_mat, gps, cutoff = 0.1, n = 5) {
pos <- lapply(gps, function(gp) {
vals <- score_mat[, gp]
cand <- names(vals)[vals > cutoff]
top <- head(cand[order(vals[cand], decreasing = TRUE)], n)
paste(top, collapse = "; ")
})
neg <- lapply(gps, function(gp) {
vals <- score_mat[, gp]
cand <- names(vals)[vals < -cutoff]
top <- head(cand[order(abs(vals[cand]), decreasing = TRUE)], n)
paste(top, collapse = "; ")
})
list(pos = unlist(pos), neg = unlist(neg))
}
gene_sig <- top_signatures(F_pm_filtered, gps)
prot_sig <- top_signatures(Protein_F_pm, gps)
supp_table <- data.frame(
GP = gp_labels,
Lineage = unlist(lineage_cats),
Cluster = unlist(cluster_cats),
Tissue = unlist(tissue_cats),
Signature_Genes_Pos = gene_sig$pos,
Signature_Genes_Neg = gene_sig$neg,
Signature_Proteins_Pos = prot_sig$pos,
Signature_Proteins_Neg = prot_sig$neg,
stringsAsFactors = FALSE
)
write.csv(
supp_table,
file = paste0(output_path, "ExtendedDataTable1_GP_summary.csv"),
row.names = FALSE
)Showing the first 20 rows (of 200); download the full table at figures/generated/ExtendedDataTable1_GP_summary.csv.
| GP | Lineage | Cluster | Tissue | Signature_Genes_Pos | Signature_Genes_Neg | Signature_Proteins_Pos | Signature_Proteins_Neg |
|---|---|---|---|---|---|---|---|
| GP1 | CD8.K; CD4.X; CD4.Y; CD4.P; Treg.P; gdT.B; gdT.C; gdT.D; gdT.P; CD8aa.C; CD8aa.D; CD8aa.E; CD8aa.P; DN.P; DP.wY; DP.wP | CT010467.1; Tmsb4x; Actb; Tmsb10; Eef1a1 | THY1.2; CD45RB; CD5; CD2; CD55.DAF | ||||
| GP2 | CD8.wM; CD4.wM; Treg.wM; gdT.E; gdT.wM; CD8aa.wM; Tz.wM; DN.wM; DP.wC; DP.wM | Arhgap15; Dock2; Mbnl1; Inpp4b; Foxp1 | Tmsb10; Tmsb4x; Fau; Pfn1; Ppia | CD45RB; THY1.2; CD44; CD24; CD27 | CD4; ICOS.CD278 | ||
| GP3 | gdT.A; gdT.B; gdT.C; gdT.D; gdT.E; gdT.F; gdT.Z; gdT.P; CD8aa.C; CD8aa.D; CD8aa.E; DP.wY | small intestine epi | Gzma; Cd7; Ccl5; Fcer1g; Tyrobp | Tmsb10; Tpt1; CT010467.1; Actb; Tmsb4x | CD38; CD8A; ITB7; CD39; CD73.5NTD | THY1.2; CD5; CD2; CD11A; CD44 | |
| GP4 | CD4.wN | Tmsb10; Fau; CT010467.1; Pfn1; Ppia | Dock2; Rack1; Arhgap15; Myh9; Ptprc | CD5; CD45RB; CD55.DAF; CD4; CD31 | |||
| GP5 | CD8.K; CD8.P; CD4.P; Treg.P; gdT.P; CD8aa.P; Tz.P; DN.P; DP.wP | H2ac8; H1f5; H1f4; H4c4; Hist1h2ap | Igfbp4; Cnn2; Limd2; Ccr7; Cd52 | CD8A; CD8B; CD73.5NTD; CD38; CD103 | CD4; CD62L | ||
| GP6 | CD8.K; CD4.T; CD4.U; CD4.X; CD4.Y; CD4.P; Treg.D; Treg.E; Treg.P; gdT.Q; gdT.R; gdT.S; gdT.T; gdT.U; gdT.V; gdT.X; gdT.Z; Tz.D | skin | S100a6; Gzmb; Lgals1; Vim; Crip1 | Tmsb10; CT010467.1; Eef1a1; Cmss1; Cdk8 | ITAM.CD11B; CD86; CD39; GR1-LY6G-LY6C1-LY6C2; GITR.CD357 | THY1.2; ICOS.CD278; KLRG1; CD5; CD31 | |
| GP7 | gdT.Z; DP.wG | CT010467.1; Ccl5; Cmss1; Cdk8; Crip1 | Igfbp4; Ccr7; Limd2; Ifi27l2a; Cd52 | THY1.2; CD45RB; CD24; CD29; CD73.5NTD | CD4; CD62L | ||
| GP8 | gdT.X; gdT.Z | skin; CNS | Fth1; S100a6; Thy1; Ctla2a; Gem | Ccl5; Gzma; Gzmb; AW112010; Tyrobp | TCRVG3; CD2; CD4; CD44; TCRGD | CD73.5NTD; CD38; CD8A; CD45RB; CD8B | |
| GP9 | CD8.wU; CD8.wV; DP.wC; DP.wE | Cdk8; Cmss1; CT010467.1; Eef2; Myh9 | Tmsb10; Fau; Tmsb4x; Cd52; Eif1 | CD45RB; CD4; CD5; CD62L; CD55.DAF | |||
| GP10 | CD8.H; CD8.I; CD8.K; CD8.wY; CD4.S; DN.F | Ccl5; Gzma; Lgals1; Nkg7; Ly6c2 | Ly6e; Ltb; Tcf7; Eef1b2; Eef1a1 | KLRG1; CD8B; CD8A; ITAX.CD11C; ICAM1 | CD4; CD62L; CD5; CD31; CD55.DAF | ||
| GP11 | CD4.X | lung; placenta; kidney | Vps37b; Gramd3; Dusp10; Junb; Ccr7 | Cd52; Ltb; Ptprcap; Mir142hg; Pfn1 | CD45RB; CD5; CD2; CD31; CD55.DAF | CD27; CD44; CD80 | |
| GP12 | CD8.C; CD8.wX; CD4.E; CD4.G; CD4.H; CD4.I; Treg.B; Tz.C; DN.A; DN.G | Izumo1r; Cdk8; Tox; Tbc1d4; Pdcd1 | Crip1; Ccl5; Satb1; Igfbp4; Selplg | FR4; CD73.5NTD; ICOS.CD278; CD38; NEUROPILIN1.CD304 | CD62L; CD45RB; CD8A; CD55.DAF; CD8B | ||
| GP13 | CD4.Y; gdT.Q; gdT.R; gdT.S; gdT.T; gdT.U; gdT.V; Tz.D | Tmem176b; Tmem176a; Il7r; CT010467.1; Cd163l1 | Ms4a4b; Inpp4b; Ctla4; Ifi27l2a; Cd2 | CD44; TCRGD; CD73.5NTD; NEUROPILIN1.CD304; THY1.2 | CD4; CD62L; CD86; SCA1; CD2 | ||
| GP14 | DP.wF; DP.wG | Cd8b1; Cd8a; Satb1; Rag1; Themis | H2-K1; Crip1; H2-D1; B2m; Eef1a1 | CD8B; CD8A; CD155.PVR; CD4; SLAM.CD150 | ITB7; SCA1; CD44; CD73.5NTD; CD55.DAF | ||
| GP15 | CD8.D; CD8.F; CD4.F; Treg.C; gdT.X; DN.G; DN.H; DP.wH | Eif5a; Npm1; Hsp90ab1; Ncl; Mif | Malat1; Tmsb4x; Itm2b; Cd52; Ypel3 | CD8B; IL2RA.CD25; CD8A; ICAM1; CD69 | CD4; CD62L; SCA1; CD5; CD31 | ||
| GP16 | CD8.F; CD8.wY; CD4.D; DN.I; DN.J | Ifi27l2a; Isg15; Ly6a; Bst2; Stat1 | Thy1 | SCA1; ICOS.CD278; CD4; CD11A; CD5 | CD38; CD8A | ||
| GP17 | CNS | Ifi27l2a; Atp5e; Tmsb4x; Ripor2; Lef1 | Actg1; Ubb; Hspa8; Eef1a1; Il31ra | CD24; SCA1; ITAM.CD11B; CD38; GR1-LY6G-LY6C1-LY6C2 | CD44; THY1.2; KLRG1; CD103; ITB7 | ||
| GP18 | Cpa3; CT010467.1; Crip1; Ccl5; Tmsb10 | Ifi27l2a; Ccr7; Ly6d; Rgcc; Limd2 | ITB7; CD27; THY1.2; CD62L; CD103 | IL2RA.CD25; CD4; FR4; CD55.DAF | |||
| GP19 | Prl8a9; Tpbpa; Psg21; Prl3b1; Creg1 | Cts6; Ccr10; Cts3; Hsd17b2; Plekhh3 | CD86; CD8B; CD4; GITR.CD357; LY49A | CD62L; THY1.2; FR4; CD55.DAF; SCA1 | |||
| GP20 | CD8.K | Cenpa; Ube2c; Cdc20; Cenpf; Arl6ip1 | H2ac8; H1f5; H1f4; Hist1h2ap; H1f3 | CD8B; CD8A; CD45RB; CD55.DAF; ITB7 |
Extended Data Table 1. One row per GP. Lineage, Cluster and Tissue list the categories (annotation_level1, annotation_level2, organ_simplified) that the GP positively predicts well – one-vs-rest AUC > 0.8 and the optimal decision threshold at or above the GP’s median loading, so high (not low) loading drives the prediction – computed on healthy non-thymocyte cells (the same population as Figure 2 / Figure 4), with the median loading taken on that same population. Signature genes / proteins give the top 5 up- and top 5 down-regulated genes and proteins by factor score, among those with |score| > 0.1 on the max|.|=1-per-GP-scaled gene and protein factor matrices.
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: Asia/Tokyo
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_1.8.7 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 stringr_1.6.0
[25] compiler_4.5.1 fs_1.6.6 Rcpp_1.1.1-1.1 pkgconfig_2.0.3
[29] later_1.4.4 digest_0.6.39 R6_2.6.1 pillar_1.11.1
[33] magrittr_2.0.5 bslib_0.9.0 tools_4.5.1 cachem_1.1.0