Skip to content

fix: debug fgsea error #167

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
Open

fix: debug fgsea error #167

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
ea06fab
fix: debug fgsea error
dlaehnemann Oct 31, 2025
d24ffb3
Merge branch 'main' into fix/debug-fgsea-stack-error
dlaehnemann Oct 31, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 2 additions & 3 deletions workflow/scripts/fgsea.R
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@ log <- file(snakemake@log[[1]], open="wt")
sink(log)
sink(log, type="message")

rlang::global_entrace()
Copy link
Contributor

@coderabbitai coderabbitai bot Oct 31, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Potential issue | 🔴 Critical

🧩 Analysis chain

Verify that rlang::global_entrace() actually resolves the C stack error.

The PR description indicates a "C stack usage is too close to the limit" error. However, rlang::global_entrace() is primarily an error tracing utility that captures better backtraces—it doesn't directly address C stack limit issues, which typically arise from deep recursion or large nested data structures.

Could you confirm that adding this line actually resolves the C stack error? If the issue persists, consider these alternatives:

  • Adjust the nproc parameter in the fgsea() call (line 52) to control parallelization
  • Increase expression depth limit with options(expressions = 50000) before the fgsea() call
  • Check if input data size can be reduced or chunked

Run the script with the changes to confirm the C stack error is resolved:

🏁 Script executed:

#!/bin/bash
# Description: Check if rlang is listed as a dependency
# This helps verify if rlang::global_entrace() will work as expected

# Search for rlang in package dependencies (e.g., DESCRIPTION, environment.yaml, requirements files)
fd -t f -e yaml -e yml -e txt --exec rg -l "rlang" {}
rg -n "library.*rlang|require.*rlang" --type r

Length of output: 0

🏁 Script executed:

#!/bin/bash
# Find project structure and R package metadata files
echo "=== Project metadata files ==="
fd -t f \( -name "DESCRIPTION" -o -name "renv.lock" -o -name ".Rprofile" -o -name "environment.yml" \)

echo -e "\n=== Listing workflow/scripts directory ==="
fd -t f --base-directory workflow/scripts

echo -e "\n=== Contents of fgsea.R (first 60 lines) ==="
head -60 workflow/scripts/fgsea.R

echo -e "\n=== Search for package dependencies on rlang ==="
rg -n "rlang|fgsea" --type r -C 2 workflow/scripts/fgsea.R | head -80

Length of output: 6701

🏁 Script executed:

#!/bin/bash
# Verify if fgsea depends on rlang or tidyverse
echo "=== Search for package dependency information ==="
# Look for any R package metadata that might list dependencies
fd -t f \( -name "DESCRIPTION" -o -name "*.lock" \) 

echo -e "\n=== Check common.R to understand what's loaded ==="
head -30 workflow/scripts/common.R

echo -e "\n=== Search for any version control or environment files ==="
fd -t f \( -name "*.lock" -o -name "*requirements*" -o -name "*.txt" \) --max-depth 2 2>/dev/null | head -20

Length of output: 1263

🌐 Web query:

rlang global_entrace C stack error fix

💡 Result:

The C-stack errors are caused by stack overflows (deep/infinitely recursive code) while rlang’s global entrace/error handlers try to capture a backtrace. Fixes:

  • Stop the recursion (best fix).
  • Temporarily disable rlang’s global entracing when running code that may overflow:
    • In your session: rlang::global_entrace(FALSE).
    • Or remove/avoid calling rlang::global_entrace() from your Rprofile. [1][2]
  • Upgrade rlang and R (rlang’s handlers use R >= 4.0 globalCallingHandlers; newer rlang releases include behavior and docs about stack-overflow handling). If you rely on handlers for diagnostics, keep rlang up to date. [1][3]
  • If you must catch stack overflows, use tryCatch/try_fetch around the risky call (note: handlers behave differently for stack overflows and R >= 4.2). See rlang docs on try_fetch and stack overflows. [4]

References:
[1] rlang::global_entrace documentation. [2] rlang::entrace docs (how entrace/handlers work). [3] rlang global_handle / handlers docs. [4] rlang docs on try_fetch and stack overflows.

rlang::global_entrace() likely worsens rather than fixes C stack overflow errors.

Based on rlang documentation, global_entrace() captures detailed error backtraces and adds error handling overhead. For C stack overflow errors (like "C stack usage 7974932 is too close to the limit"), this approach is counterproductive because it increases stack consumption during error processing.

The recommended fixes for C stack overflows are:

  • Disable global_entrace() rather than enable it
  • Reduce nproc parameter in the fgsea() call (line 48) to decrease parallelization overhead
  • Adjust eps parameter to affect convergence behavior
  • Reduce input data size if possible

Please verify that adding this line actually resolves the reported C stack error. If the error persists, consider removing global_entrace() and implementing one of the alternatives above.

🤖 Prompt for AI Agents 
In workflow/scripts/fgsea.R around line 5, the call to rlang::global_entrace()
is likely increasing stack usage and can worsen C stack overflow errors; remove
or comment out this line so backtrace capture is disabled, then in the fgsea()
invocation around line 48 reduce the nproc value (e.g., lower or set to 1) to
limit parallel threads, consider increasing the eps parameter to relax
convergence or otherwise tune it, and if feasible reduce the input data size;
after these changes run the failing analysis to confirm the C stack error is
resolved and only re-enable advanced error tracing if the overflow no longer
occurs.


library("fgsea")
library(snakemake@params[["bioc_species_pkg"]], character.only = TRUE)
# avoid this error: https://github.com/ctlab/fgsea/issues/98
Expand Down Expand Up @@ -43,9 +45,6 @@ rank_ties <- enframe(ranked_genes) %>%
dplyr::select(ext_gene = name, !!signed_pi := value)
write_tsv(rank_ties, snakemake@output[["rank_ties"]])

print(gene_sets)
print(ranked_genes)

fgsea_res <- fgsea(pathways = gene_sets,
stats = ranked_genes,
minSize=10,
Expand Down
Loading