Skip to content

fix CTWAS again #569

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

Merged
danielnachun merged 1 commit into from
Aug 8, 2026
Merged

fix CTWAS again #569

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
19 changes: 13 additions & 6 deletions R/ctwasPipeline.R
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -464,8 +464,11 @@ estCtwasParam <- function(inputs,
if (fallbackToPrefit) {
message("estCtwasParam: accurate EM unusable (",
conditionMessage(e), "); falling back to prefit estimates.")
# Run the prefit EM to CONVERGENCE (full `niter`), not the rough
# niter_prefit warm-up count -- this is now the final prior, so an
# under-iterated fit would depress downstream gene PIPs.
.ctwasFitPrefitEm(regionData,
niterPrefit = as.integer(niterPrefit),
niter = as.integer(niter),
groupPriorVarStructure = groupPriorVarStructure,
thin = thin,
ncore = as.integer(ncore),
Expand Down Expand Up @@ -738,12 +741,16 @@ mergeCtwasBoundaryRegions <- function(finemapResult,
# param list shaped like ctwas::est_param normally produces. Used as
# the fallback path when est_param's accurate EM diverges to NaN on
# toy / underpowered data (matches the legacy ctwas_2 workaround).
# Calls ctwas's internal `fit_EM` (via getFromNamespace) with
# niter = niter_prefit, then applies the same thin-adjustment to the
# SNP group_prior that est_param applies. p_single_effect is left as
# Calls ctwas's internal `fit_EM` (via getFromNamespace) for `niter`
# iterations -- run to convergence, because this fallback prior is the
# FINAL estimate (the accurate EM never ran), not a warm-up. `niter` here
# is the caller's full accurate-EM count, NOT niter_prefit; running only
# niter_prefit (a rough warm-up, e.g. 3) leaves the prior under-converged
# and depresses downstream gene PIPs. Then applies the same thin-adjustment
# to the SNP group_prior that est_param applies. p_single_effect is left as
# NA since the accurate EM never ran.
# @noRd
.ctwasFitPrefitEm <- function(region_data, niterPrefit,
.ctwasFitPrefitEm <- function(region_data, niter,
groupPriorVarStructure, thin, ncore,
extra = list()) {
fitEm <- getFromNamespace("fit_EM", "ctwas")
Expand All @@ -767,7 +774,7 @@ mergeCtwasBoundaryRegions <- function(finemapResult,
stop("No regions selected!")
fitArgs <- list(
region_data = fitRegionData,
niter = as.integer(niterPrefit),
niter = as.integer(niter),
group_prior_var_structure = groupPriorVarStructure,
ncore = as.integer(ncore))
prefit <- .ctwasInvoke(fitEm, fitArgs, extra)
Expand Down
2 changes: 1 addition & 1 deletion tests/testthat/test_ctwasPipeline.R
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1342,7 +1342,7 @@ test_that("(real ctwas) prefit fallback skips a degenerate region fit_EM would r
region_data <- readRDS(test_path("test_data", "ctwas_region_data_degenerate.rds"))
expect_length(region_data, 2L)
res <- .ctwasFitPrefitEm(
region_data, niterPrefit = 3L,
region_data, niter = 3L,
groupPriorVarStructure = "shared_all", thin = 1, ncore = 1L)
# the prefit EM ran on the valid region only and returns finite real group priors
expect_true("SNP" %in% names(res$group_prior))
Expand Down