diff --git a/R/GenotypeHandle.R b/R/GenotypeHandle.R
index 43baff5f..5f94f4d6 100644
--- a/R/GenotypeHandle.R
+++ b/R/GenotypeHandle.R
@@ -130,6 +130,16 @@ setMethod("show", "GenotypeHandle", function(object) {
 #'   \code{#chr,path} meta file or a named character vector
 #'   (names = chromosomes, values = payload paths/prefixes). Optionally pass
 #'   \code{format} via \code{...} to force a single backend for every shard.
+#' @param chroms Optional character vector of chromosomes. With \code{genoMeta}
+#'   (a sharded, one-file-per-chromosome panel) only the shards whose
+#'   chromosome is listed are read, skipping the rest -- an I/O optimisation
+#'   when the panel is genome-wide but only a few chromosomes are needed.
+#'   Chromosome labels are compared canonically (\code{"chr1"}/\code{"1"} match,
+#'   \code{23}/\code{X} etc.). If none of the requested chromosomes are present
+#'   in the panel, every shard is read (so the caller's own absence check can
+#'   report the mismatch). Only meaningful with \code{genoMeta}; supplying it
+#'   with any other source is an error (a single-file panel has no per-chromosome
+#'   shards to skip).
 #' @param ... Additional arguments forwarded to the format-specific reader.
 #' @return A \code{GenotypeHandle} object.
 #' @export
@@ -138,7 +148,7 @@ GenotypeHandle <- function(path = NULL,
                            bed = NULL, bim = NULL, fam = NULL,
                            pgen = NULL, pvar = NULL, psam = NULL,
                            ldMeta = NULL, region = NULL,
-                           genoMeta = NULL,
+                           genoMeta = NULL, chroms = NULL,
                            ...) {
   bedTrioGiven <- !is.null(bed) || !is.null(bim) || !is.null(fam)
   bedTrioComplete <- !is.null(bed) && !is.null(bim) && !is.null(fam)
@@ -173,6 +183,10 @@ GenotypeHandle <- function(path = NULL,
          "bed/bim/fam triplet, the pgen/pvar/psam triplet, `ldMeta`, or ",
          "`genoMeta` must be specified (got ", nSources, ").")
   }
+  if (!is.null(chroms) && !sources[["genoMeta"]]) {
+    stop("`chroms` restricts which per-chromosome shards are read and is only ",
+         "supported with `genoMeta` (a single-file panel has no shards to skip).")
+  }
 
   if (sources[["path"]]) {
     return(readGenotypes(path, ...))
@@ -193,7 +207,7 @@ GenotypeHandle <- function(path = NULL,
     return(.genotypeHandleFromLdMeta(ldMeta, region, ...))
   }
   if (sources[["genoMeta"]]) {
-    return(.genotypeHandleFromChromMeta(genoMeta, ...))
+    return(.genotypeHandleFromChromMeta(genoMeta, chroms = chroms, ...))
   }
 }
 
@@ -393,14 +407,28 @@ GenotypeHandle <- function(path = NULL,
 # metadata via the existing single-file readers, validates a single shared
 # format and identical sample IDs (same order, required for cross-shard
 # cbind), and row-binds the per-shard snpInfo into one global index space.
+# `chroms` (optional) restricts the read to the shards for those chromosomes:
+# the other per-chromosome files are never opened, which is the I/O win when a
+# genome-wide panel backs summary statistics on only a few chromosomes.
 #' @keywords internal
-.genotypeHandleFromChromMeta <- function(genoMeta, ...) {
+.genotypeHandleFromChromMeta <- function(genoMeta, chroms = NULL, ...) {
   dots   <- list(...)
   format <- dots$format
   parsed <- .parseChromMeta(genoMeta)
   if (nrow(parsed) == 0L)
     stop("GenotypeHandle(genoMeta): no chromosomes found in the meta input.")
 
+  # Keep only the shards whose (meta-declared) chromosome was requested, so the
+  # rest are never read. If nothing matches -- the requested chromosomes are
+  # absent from the panel -- fall back to reading every shard, leaving the
+  # caller's own containment/absence check to produce the usual diagnostic
+  # instead of a confusing empty-handle error here.
+  if (!is.null(chroms)) {
+    keep <- canonChrom(as.character(parsed$chrom)) %in%
+            canonChrom(as.character(chroms))
+    if (any(keep)) parsed <- parsed[keep, , drop = FALSE]
+  }
+
   shards <- lapply(parsed$path, .resolveGenotypeShard, format = format)
 
   formats <- vapply(shards, function(h) h@format, character(1))
diff --git a/R/colocboostPipeline.R b/R/colocboostPipeline.R
index fb34a93b..4b2f0c2e 100644
--- a/R/colocboostPipeline.R
+++ b/R/colocboostPipeline.R
@@ -81,6 +81,14 @@
 #'   skipped. \code{0} (default) disables it; a negative value uses
 #'   \code{3 / n_variants}. (Summary-statistic skipping is handled upstream by
 #'   \code{\link{summaryStatsQc}}'s own \code{pipCutoffToSkip}.)
+#' @param absZCutoffToSkip,bfCutoffToSkip,logBfCutoffToSkip Alternative
+#'   individual-level pre-filter metrics used in place of
+#'   \code{pipCutoffToSkip}: drop an outcome unless its maximum marginal
+#'   \code{|z|} (\code{absZCutoffToSkip}), or its maximum per-variant
+#'   single-effect Bayes factor (\code{bfCutoffToSkip}) / log Bayes factor
+#'   (\code{logBfCutoffToSkip}) from the \code{L = 1} fit, exceeds the cutoff.
+#'   Scalars, each defaulting to 0 (off). Exactly one screening metric may be
+#'   enabled: setting any of these requires \code{pipCutoffToSkip = 0}.
 #' @param alleleFlip Logical, default \code{TRUE}. When TRUE, harmonize variants
 #'   across the individual X, sumstats, and LD by (chrom, pos) with ref/alt
 #'   swaps recognized (flipping z / residualized dosage / LD to a shared
@@ -160,10 +168,13 @@ setGeneric("colocboostPipeline",
   invisible(NULL)
 }
 
-# Resolve the per-context pipCutoffToSkip from either a scalar (applies to
-# every context) or a named vector keyed by context. Default 0 (no skip).
+# Resolve the per-context screen spec. The `pipCutoffToSkip` channel carries
+# EITHER a resolved screen object (list(metric, cutoff) for absZ/bf/logBf/pip --
+# applies uniformly to every context) OR the legacy PIP cutoff as a scalar (all
+# contexts) or a named vector keyed by context. Default 0 (no screen).
 .cbResolveCutoff <- function(pipCutoffToSkip, ctx) {
   if (is.null(pipCutoffToSkip) || length(pipCutoffToSkip) == 0L) return(0)
+  if (is.list(pipCutoffToSkip)) return(pipCutoffToSkip)   # uniform screen object
   if (!is.null(names(pipCutoffToSkip))) {
     if (ctx %in% names(pipCutoffToSkip)) return(pipCutoffToSkip[[ctx]])
     return(0)
@@ -171,18 +182,35 @@ setGeneric("colocboostPipeline",
   pipCutoffToSkip[[1L]]
 }
 
+# Combine the four colocboost screen cutoffs into a single spec to thread
+# through the pipCutoffToSkip channel: a resolved screen object when a new
+# metric (absZ / bf / logBf) is set, otherwise the (possibly context-named)
+# legacy pipCutoffToSkip. Enforces one screening metric at a time.
+.cbScreenSpec <- function(pipCutoffToSkip, absZCutoffToSkip,
+                          bfCutoffToSkip, logBfCutoffToSkip) {
+  newScreen <- .resolveScreenMetric(0, absZCutoffToSkip,
+                                    bfCutoffToSkip, logBfCutoffToSkip)
+  pipOn <- !is.null(pipCutoffToSkip) && length(pipCutoffToSkip) > 0L &&
+           any(as.numeric(pipCutoffToSkip) != 0, na.rm = TRUE)
+  if (!is.null(newScreen) && pipOn)
+    stop("colocboostPipeline: only one signal screen may be enabled at a time; ",
+         "unset pipCutoffToSkip to use absZCutoffToSkip / bfCutoffToSkip / ",
+         "logBfCutoffToSkip.")
+  if (!is.null(newScreen)) newScreen else pipCutoffToSkip
+}
+
 # Per-outcome single-trait skip (ports the legacy qc_individual_data
 # pip_cutoff_to_skip): for each outcome column of Y, fit a single-effect
 # SuSiE (L = 1, max_iter = 100) on (X, Y[, j]) and keep the outcome only if
 # any variant's PIP exceeds the cutoff. A cutoff < 0 means 3 / n_variants.
 # Returns the retained Y (NULL when no outcome clears the threshold).
-.cbPipSkipOutcomes <- function(X, Y, cutoff) {
-  if (is.null(cutoff) || is.na(cutoff) || cutoff == 0) return(Y)
+.cbPipSkipOutcomes <- function(X, Y, spec) {
+  if (is.null(.asScreen(spec))) return(Y)
   # Single-effect screen per outcome, sharing the L = 1 SuSiE pre-screen
   # (.fmSerScreen) with the fine-mapping pipeline. fallback = FALSE: an outcome
   # that cannot be screened (too few samples / fit failure) is dropped.
   keep <- vapply(seq_len(ncol(Y)),
-                 function(j) .fmSerScreen(X, Y[, j], cutoff, fallback = FALSE),
+                 function(j) .fmSerScreen(X, Y[, j], spec, fallback = FALSE),
                  logical(1L))
   if (!any(keep)) return(NULL)
   Y[, keep, drop = FALSE]
@@ -251,11 +279,11 @@ setGeneric("colocboostPipeline",
     X <- X[common, , drop = FALSE]
     Y <- Y[common, , drop = FALSE]
     cutoffCtx <- .cbResolveCutoff(pipCutoffToSkip, ctx)
-    if (!is.null(cutoffCtx) && !is.na(cutoffCtx) && cutoffCtx != 0) {
+    if (!is.null(.asScreen(cutoffCtx))) {
       Y <- .cbPipSkipOutcomes(X, Y, cutoffCtx)
       if (is.null(Y) || ncol(Y) == 0L) {
         message("colocboostPipeline: skipping context '", ctx,
-                "' (no outcome cleared pipCutoffToSkip = ", cutoffCtx, ").")
+                "' (no outcome cleared the signal screen).")
         next
       }
     }
@@ -699,9 +727,14 @@ setMethod("colocboostPipeline", "QtlDataset",
            separateGwas = FALSE,
            samples = NULL,
            pipCutoffToSkip = 0,
+           absZCutoffToSkip = 0,
+           bfCutoffToSkip = 0,
+           logBfCutoffToSkip = 0,
            alleleFlip = TRUE,
            ...) {
     dotArgs <- list(...)
+    screenSpec <- .cbScreenSpec(pipCutoffToSkip, absZCutoffToSkip,
+                                bfCutoffToSkip, logBfCutoffToSkip)
     indBundle <- .cbIndividualBundle(
       qtlData,
       contexts     = contexts,
@@ -709,7 +742,7 @@ setMethod("colocboostPipeline", "QtlDataset",
       region       = region,
       cisWindow    = cisWindow,
       samples      = samples,
-      pipCutoffToSkip = pipCutoffToSkip)
+      pipCutoffToSkip = screenSpec)
     .cbDriver(indBundle, qtlPairs = list(), gwasSumStats,
               xqtlColoc, jointGwas, separateGwas,
               focalTrait, dotArgs, alleleFlip = alleleFlip)
@@ -755,9 +788,14 @@ setMethod("colocboostPipeline", "MultiStudyQtlDataset",
            separateGwas = FALSE,
            samples = NULL,
            pipCutoffToSkip = 0,
+           absZCutoffToSkip = 0,
+           bfCutoffToSkip = 0,
+           logBfCutoffToSkip = 0,
            alleleFlip = TRUE,
            ...) {
     dotArgs <- list(...)
+    screenSpec <- .cbScreenSpec(pipCutoffToSkip, absZCutoffToSkip,
+                                bfCutoffToSkip, logBfCutoffToSkip)
 
     # Aggregate the individual-level bundles across all QtlDataset
     # members. Per-study trait names are prefixed with "{study}:" so
@@ -777,7 +815,7 @@ setMethod("colocboostPipeline", "MultiStudyQtlDataset",
         region       = region,
         cisWindow    = cisWindow,
         samples      = samples,
-        pipCutoffToSkip = pipCutoffToSkip)
+        pipCutoffToSkip = screenSpec)
       if (is.null(sub)) next
       xOffset <- length(combinedX)
       yOffset <- length(combinedY)
diff --git a/R/ctwasPipeline.R b/R/ctwasPipeline.R
index 75ebc7be..e1bfdc51 100644
--- a/R/ctwasPipeline.R
+++ b/R/ctwasPipeline.R
@@ -747,8 +747,26 @@ mergeCtwasBoundaryRegions <- function(finemapResult,
                               groupPriorVarStructure, thin, ncore,
                               extra = list()) {
   fitEm <- getFromNamespace("fit_EM", "ctwas")
+  # Mirror ctwas::est_param's degenerate-region skip before fitting. est_param
+  # drops regions with fewer than `min_var` total variables or fewer than
+  # `min_gene` genes -- whose per-region `sid` is unset -- and runs its prefit
+  # fit_EM only on the survivors. This prefit-only fallback bypasses est_param's
+  # p(single effect) selection gate, but it must apply the SAME filter, or
+  # ctwas::fit_EM errors inside extract_region_data ("regiondata$sid ... target
+  # is NULL") on a skipped region. Honor min_var / min_gene forwarded via `...`.
+  minVar  <- if (!is.null(extra$min_var))  as.integer(extra$min_var)  else 2L
+  minGene <- if (!is.null(extra$min_gene)) as.integer(extra$min_gene) else 1L
+  allRegionIds <- names(region_data)
+  nGid <- vapply(region_data, function(x) length(x$gid), integer(1L))
+  nSid <- vapply(region_data, function(x) length(x$sid), integer(1L))
+  keep <- rep(TRUE, length(region_data))
+  if (minVar  > 0L) keep <- keep & (nSid + nGid) >= minVar
+  if (minGene > 0L) keep <- keep & nGid >= minGene
+  fitRegionData <- region_data[keep]
+  if (length(fitRegionData) == 0L)
+    stop("No regions selected!")
   fitArgs <- list(
-    region_data               = region_data,
+    region_data               = fitRegionData,
     niter                     = as.integer(niterPrefit),
     group_prior_var_structure = groupPriorVarStructure,
     ncore                     = as.integer(ncore))
@@ -771,7 +789,7 @@ mergeCtwasBoundaryRegions <- function(finemapResult,
     group_prior_var_structure = groupPriorVarStructure,
     group_size                = groupSize,
     p_single_effect           = data.frame(
-      region_id        = names(region_data),
+      region_id        = allRegionIds,
       p_single_effect  = NA_real_,
       stringsAsFactors = FALSE))
 }
diff --git a/R/fineMappingPipeline.R b/R/fineMappingPipeline.R
index 471c3aef..6f2bce1f 100644
--- a/R/fineMappingPipeline.R
+++ b/R/fineMappingPipeline.R
@@ -197,6 +197,14 @@
 #'   summary-statistics analog lives in \code{summaryStatsQc()}. \code{0}
 #'   (default) disables the screen; a negative value uses the adaptive
 #'   \code{3 / nVariants} threshold.
+#' @param absZCutoffToSkip,bfCutoffToSkip,logBfCutoffToSkip Numeric (length 1).
+#'   Alternative individual-level pre-screen metrics, in place of
+#'   \code{pipCutoffToSkip}: skip a block unless its maximum marginal
+#'   \code{|z|} (\code{absZCutoffToSkip}), or its maximum per-variant
+#'   single-effect Bayes factor (\code{bfCutoffToSkip}) / log Bayes factor
+#'   (\code{logBfCutoffToSkip}) from the \code{L = 1} fit, exceeds the cutoff.
+#'   Each defaults to 0 (off). Exactly one of the four \code{*CutoffToSkip}
+#'   arguments may be non-zero (one screening metric at a time).
 #' @param usePCA Logical (length 1). \code{QtlDataset} only. When
 #'   \code{TRUE} (default \code{FALSE}), each multi-trait context's
 #'   PCA-reduced phenotype is fine-mapped with univariate SuSiE on its
@@ -1002,14 +1010,25 @@ combineFineMappingResults <- function(..., ldSketch = NULL) {
   scores
 }
 
-# Single-effect (SER) pre-screen, individual-level. Fits susie with L = 1 on a
-# residualized (X, y) block and reports whether any PIP clears `cutoff` -- i.e.
-# whether the block shows any potentially significant variant worth a full fit.
-# Ports the deleted multivariate_pipeline.R `skipConditions` / susie_twas
-# `pip_cutoff_to_skip` logic (the individual-level analog of the sumstat-path
-# `.applyPipScreen`):
-#   * `cutoff == 0` (or NULL/non-scalar) disables the screen -> always keep.
-#   * `cutoff < 0` uses the adaptive 3 / nVariants threshold.
+# Per-column marginal-association z-scores of y on each column of X (univariate
+# regression z = betahat / sebetahat), used by the individual-level absZ screen.
+# @noRd
+.marginalZ <- function(X, y) {
+  ur <- susieR::univariate_regression(X, y)
+  ur$betahat / ur$sebetahat
+}
+
+# Single-effect (SER) pre-screen, individual-level. Reports whether a
+# residualized (X, y) block shows a strong enough signal (by the chosen metric)
+# to be worth a full fit. `screen` is a screen spec (see .asScreen): a legacy
+# PIP cutoff (numeric scalar, 0 = off) OR a resolved list(metric, cutoff) for
+# one of pip / absZ / bf / logBf. Ports the deleted multivariate_pipeline.R
+# `skipConditions` / susie_twas `pip_cutoff_to_skip` logic (the individual-level
+# analog of the sumstat-path `.applyEntryScreen`):
+#   * no screen (NULL / 0 / non-scalar numeric) -> always keep.
+#   * pip cutoff < 0 uses the adaptive 3 / nVariants threshold.
+#   * absZ needs no susie fit; pip/bf/logBf fit susie L = 1 once (its
+#     $lbf_variable gives the per-variant logBF for bf/logBf, $pip for pip).
 #   * NA entries of `y` are dropped before fitting.
 # The screen is advisory: too few samples/variants or a fit failure returns
 # `fallback` -- TRUE (default) keeps the block rather than discard a potentially
@@ -1017,26 +1036,39 @@ combineFineMappingResults <- function(..., ldSketch = NULL) {
 # outcome it cannot screen. This is the single L = 1 SuSiE pre-screen shared by
 # .fmSerScreenColumns (joint) and .cbPipSkipOutcomes (colocboost).
 # @noRd
-.fmSerScreen <- function(X, y, cutoff, fallback = TRUE) {
-  if (is.null(cutoff) || length(cutoff) != 1L || is.na(cutoff) || cutoff == 0)
-    return(TRUE)
+.fmSerScreen <- function(X, y, screen, fallback = TRUE) {
+  scr <- .asScreen(screen)
+  if (is.null(scr)) return(TRUE)
   ok <- !is.na(y)
   if (sum(ok) < 2L || ncol(X) < 1L) return(fallback)
   Xs <- X[ok, , drop = FALSE]
   if (!is.double(Xs)) storage.mode(Xs) <- "double"   # susieR needs double X
-  thr <- if (cutoff < 0) 3 / ncol(Xs) else cutoff
-  pip <- tryCatch(suppressMessages(susieR::susie(Xs, y[ok], L = 1L))$pip,
+  ys <- y[ok]
+  metric <- scr$metric; cutoff <- scr$cutoff
+  if (metric == "absZ") {
+    z <- tryCatch(.marginalZ(Xs, ys), error = function(e) NULL)
+    if (is.null(z)) return(fallback)
+    return(any(abs(z) > cutoff, na.rm = TRUE))
+  }
+  fit <- tryCatch(suppressMessages(susieR::susie(Xs, ys, L = 1L)),
                   error = function(e) NULL)
-  if (is.null(pip)) return(fallback)
-  any(pip > thr, na.rm = TRUE)
+  if (is.null(fit)) return(fallback)
+  if (metric == "pip") {
+    thr <- if (cutoff < 0) 3 / ncol(Xs) else cutoff
+    return(any(fit$pip > thr, na.rm = TRUE))
+  }
+  maxLbf <- suppressWarnings(max(as.numeric(fit$lbf_variable), na.rm = TRUE))
+  if (!is.finite(maxLbf)) return(fallback)
+  # bf: cutoff on the raw BF scale -> compare in log space; logBf: log scale.
+  maxLbf > (if (metric == "bf") log(cutoff) else cutoff)
 }
 
-# Is the SER pre-screen enabled? Only a finite, non-zero scalar activates it;
-# this gates the extra screening extraction so the default (cutoff 0) costs
-# nothing.
+# Is a signal screen enabled? Any spec that .asScreen resolves to a screen
+# object (a non-zero pip cutoff or a resolved metric) activates it; this gates
+# the extra screening extraction so the default (no screen) costs nothing.
 # @noRd
-.fmScreenActive <- function(cutoff) {
-  !is.null(cutoff) && length(cutoff) == 1L && !is.na(cutoff) && cutoff != 0
+.fmScreenActive <- function(screen) {
+  !is.null(.asScreen(screen))
 }
 
 # Per-condition SER pre-screen for a joint (multi-context / multi-trait) fit:
@@ -1045,9 +1077,9 @@ combineFineMappingResults <- function(..., ldSketch = NULL) {
 # and a port of the deleted `skipConditions`: callers drop the FALSE columns
 # (null contexts / traits) before the joint mvSuSiE fit.
 # @noRd
-.fmSerScreenColumns <- function(X, Y, cutoff) {
+.fmSerScreenColumns <- function(X, Y, screen) {
   vapply(seq_len(ncol(Y)),
-         function(j) .fmSerScreen(X, Y[, j], cutoff),
+         function(j) .fmSerScreen(X, Y[, j], screen),
          logical(1L))
 }
 
@@ -1374,6 +1406,9 @@ setMethod("fineMappingPipeline", "QtlDataset",
            cvThreads          = 1,
            samplePartition    = NULL,
            pipCutoffToSkip    = 0,
+           absZCutoffToSkip   = 0,
+           bfCutoffToSkip     = 0,
+           logBfCutoffToSkip  = 0,
            usePCA             = FALSE,
            nPCs               = 10L,
            seed               = NULL,
@@ -1392,6 +1427,10 @@ setMethod("fineMappingPipeline", "QtlDataset",
            ...) {
     naAction <- match.arg(naAction)
     if (!is.null(seed)) set.seed(as.integer(seed))
+    # Resolve the single active signal screen (pip / absZ / bf / logBf) once and
+    # thread the resolved spec through the existing pipCutoffToSkip channels.
+    screen <- .resolveScreenMetric(pipCutoffToSkip, absZCutoffToSkip,
+                                   bfCutoffToSkip, logBfCutoffToSkip)
     # Apply any per-call filter overrides to a validated copy of the dataset
     # (replaces the construct-time slot values for this call only).
     data <- .qtlApplyFilterOverrides(data, mafCutoff, macCutoff, xvarCutoff,
@@ -1426,7 +1465,7 @@ setMethod("fineMappingPipeline", "QtlDataset",
         twasWeights = twasWeights,
         dataDrivenPriorWeightsCutoff = dataDrivenPriorWeightsCutoff,
         cvFolds = cvFolds, cvThreads = cvThreads, samplePartition = samplePartition,
-        pipCutoffToSkip = pipCutoffToSkip,
+        pipCutoffToSkip = screen,
         fineMappingResult = fineMappingResult,
         fullFit = fullFit, fullFitAlphaOnly = fullFitAlphaOnly,
         includeAllCs = includeAllCs)
@@ -1547,12 +1586,12 @@ setMethod("fineMappingPipeline", "QtlDataset",
             X <- X[common, , drop = FALSE]
             y <- Y[common, , drop = FALSE]
             if (ncol(y) > 1L) y <- y[, 1L, drop = TRUE] else y <- drop(y)
-            # SER pre-screen: skip this block when a single-effect fit finds no
-            # PIP above pipCutoffToSkip (no potentially significant variant).
-            if (!.fmSerScreen(X, y, pipCutoffToSkip)) {
+            # SER pre-screen: skip this block when the chosen signal screen
+            # finds no strong-enough variant (no potentially significant signal).
+            if (!.fmSerScreen(X, y, screen)) {
               if (verbose >= 1)
                 message(sprintf(
-                  "Skipping (context='%s', trait='%s'): SER pre-screen found no PIP above pipCutoffToSkip.",
+                  "Skipping (context='%s', trait='%s'): SER pre-screen found no signal above the cutoff.",
                   ctx, tid))
               return(list())
             }
@@ -1607,7 +1646,7 @@ setMethod("fineMappingPipeline", "QtlDataset",
             common <- intersect(rownames(X), names(pcY))
             if (length(common) < 2L) return(list())
             Xb <- X[common, , drop = FALSE]
-            if (!.fmSerScreen(Xb, pcY[common], pipCutoffToSkip)) return(list())
+            if (!.fmSerScreen(Xb, pcY[common], screen)) return(list())
             afVec <- .fmAfForX(data, Xb, traitId = traits, region = rg,
                                cisWindow = cisWindow)
             .fmFitXBlock(Xb, pcY[common], "susie", FALSE, coverage,
@@ -1643,7 +1682,7 @@ setMethod("fineMappingPipeline", "QtlDataset",
         twasWeights = twasWeights,
         dataDrivenPriorWeightsCutoff = dataDrivenPriorWeightsCutoff,
         cvFolds = cvFolds, cvThreads = cvThreads, samplePartition = samplePartition,
-        pipCutoffToSkip = pipCutoffToSkip,
+        pipCutoffToSkip = screen,
         fineMappingResult = fineMappingResult,
         fullFit = fullFit, fullFitAlphaOnly = fullFitAlphaOnly,
         includeAllCs = includeAllCs)
@@ -1698,6 +1737,8 @@ setMethod("fineMappingPipeline", "QtlDataset",
     minAbsCorr = cfg$minAbsCorr, fineMappingResult = cfg$fineMappingResult,
     cvFolds = cfg$cvFolds, cvThreads = cfg$cvThreads,
     samplePartition = cfg$samplePartition, pipCutoffToSkip = cfg$pipCutoffToSkip,
+    absZCutoffToSkip = cfg$absZCutoffToSkip, bfCutoffToSkip = cfg$bfCutoffToSkip,
+    logBfCutoffToSkip = cfg$logBfCutoffToSkip,
     seed = cfg$seed, naAction = cfg$naAction, verbose = cfg$verbose), cfg$dotArgs))
 }
 
@@ -1739,6 +1780,9 @@ setMethod("fineMappingPipeline", "MultiStudyQtlDataset",
            cvThreads          = 1,
            samplePartition    = NULL,
            pipCutoffToSkip    = 0,
+           absZCutoffToSkip   = 0,
+           bfCutoffToSkip     = 0,
+           logBfCutoffToSkip  = 0,
            seed               = NULL,
            naAction           = c("drop", "impute"),
            verbose            = 1,
@@ -1800,6 +1844,8 @@ setMethod("fineMappingPipeline", "MultiStudyQtlDataset",
                 minAbsCorr = minAbsCorr, fineMappingResult = fineMappingResult,
                 cvFolds = cvFolds, cvThreads = cvThreads,
                 samplePartition = samplePartition, pipCutoffToSkip = pipCutoffToSkip,
+                absZCutoffToSkip = absZCutoffToSkip, bfCutoffToSkip = bfCutoffToSkip,
+                logBfCutoffToSkip = logBfCutoffToSkip,
                 seed = seed, naAction = naAction, verbose = verbose, dotArgs = dotArgs)
     .multiStudyPipelineDriver(
       data, jointResult, .fmPerStudy, .fmSumStats, cfg,
diff --git a/R/manifestLoaders.R b/R/manifestLoaders.R
index 8d1f55d4..c7ca8648 100644
--- a/R/manifestLoaders.R
+++ b/R/manifestLoaders.R
@@ -142,11 +142,14 @@ NULL
 # Turn an ldSketch specification into a GenotypeHandle. Accepts a prebuilt
 # handle, a named chrom -> path vector (genoMeta), or a single string that is
 # either a genotype file/prefix or a chrom-sharded genoMeta meta file.
-.resolveLdSketch <- function(spec) {
+# `chroms` (optional) restricts a chrom-sharded panel to those chromosomes so
+# the other per-chromosome shards are never read; it is ignored for a single
+# genotype file (which has no shards to skip).
+.resolveLdSketch <- function(spec, chroms = NULL) {
   if (is.null(spec)) return(NULL)
   if (methods::is(spec, "GenotypeHandle")) return(spec)
   if (is.character(spec) && !is.null(names(spec))) {
-    return(GenotypeHandle(genoMeta = spec))
+    return(GenotypeHandle(genoMeta = spec, chroms = chroms))
   }
   if (is.character(spec) && length(spec) == 1L) {
     lower <- tolower(spec)
@@ -156,13 +159,18 @@ NULL
       file.exists(paste0(spec, ".bed")) || file.exists(paste0(spec, ".pgen"))
     if (looksLikeGenotype) return(.detectGenotypeFormat(spec))
     # Otherwise treat it as a chrom-sharded genoMeta meta file.
-    return(GenotypeHandle(genoMeta = spec))
+    return(GenotypeHandle(genoMeta = spec, chroms = chroms))
   }
   stop("`ldSketch` must be a GenotypeHandle, a genotype path/prefix, or a ",
        "genoMeta spec (named chrom->path vector or meta-file path).")
 }
 
-# Resolve the LD sketch from the (argument, ldSketchPath column) pair.
+# Resolve the LD sketch SPEC from the (argument, ldSketchPath column) pair,
+# without reading any genotype metadata yet. Returns a prebuilt GenotypeHandle
+# (when the caller passed one), or the reconciled spec (a path/prefix or a
+# genoMeta path/named-vector) for later materialisation. Deferring the read to
+# `.materializeLdSketch` lets the loader first learn which chromosomes the
+# summary statistics cover and skip the panel's other per-chromosome shards.
 .resolveLdSketchInput <- function(df, ldSketch, base) {
   if (methods::is(ldSketch, "GenotypeHandle")) return(ldSketch)
   colSpec <- NULL
@@ -181,7 +189,28 @@ NULL
   if (is.null(spec)) {
     stop("`ldSketch` must be provided as an argument or an `ldSketchPath` column.")
   }
-  .resolveLdSketch(spec)
+  spec
+}
+
+# Materialise a resolved ldSketch spec into a GenotypeHandle, reading only the
+# shards for `chroms` when the spec is a chrom-sharded panel. A spec that is
+# already a GenotypeHandle (the caller passed a prebuilt handle) is returned
+# unchanged: its snpInfo is already in memory, so there is no read to restrict.
+.materializeLdSketch <- function(spec, chroms) {
+  if (is.null(spec)) return(NULL)
+  if (methods::is(spec, "GenotypeHandle")) return(spec)
+  .resolveLdSketch(spec, chroms = chroms)
+}
+
+# Canonical chromosomes present across a list of sumstats entry GRanges. NULL /
+# empty entries contribute nothing; NA seqnames are dropped. Always returns a
+# character vector (character(0) when nothing is present, never NULL).
+.entriesChroms <- function(entries) {
+  ch <- as.character(unlist(lapply(entries, function(gr) {
+    if (is.null(gr) || length(gr) == 0L) return(character(0))
+    canonChrom(as.character(GenomicRanges::seqnames(gr)))
+  }), use.names = FALSE))
+  unique(ch[!is.na(ch)])
 }
 
 # =============================================================================
@@ -907,7 +936,7 @@ loadGwasSumStatsFromManifest <- function(manifest, genome = NULL,
     stop("GwasSumStats manifest `study` values must be unique.")
   }
   genome <- .reconcileScalar(df$genome, genome, "genome")
-  ldSketch <- .resolveLdSketchInput(df, ldSketch, base)
+  ldSketchSpec <- .resolveLdSketchInput(df, ldSketch, base)
 
   # Study-level sample-size scalars (from the manifest). When a row carries a
   # usable scalar (n_case + n_control, or n_sample), the sumstats file need not
@@ -925,13 +954,21 @@ loadGwasSumStatsFromManifest <- function(manifest, genome = NULL,
     } else {
       columnMapping
     }
-    gr <- .loadSumStatsEntry(.resolveRel(as.character(df$sumStatsPath[[i]]), base),
-                             region, mapping, sampleSelect, formatMapping, label,
-                             allowNoN = .manifestHasStudyScalar(i, nCaseCol, nControlCol, nSampleCol))
-    .checkLdContainment(ldSketch, gr, minLdOverlapWarn, label)
-    gr
+    .loadSumStatsEntry(.resolveRel(as.character(df$sumStatsPath[[i]]), base),
+                       region, mapping, sampleSelect, formatMapping, label,
+                       allowNoN = .manifestHasStudyScalar(i, nCaseCol, nControlCol, nSampleCol))
   })
 
+  # Materialise the LD sketch reading only the chromosomes the summary stats
+  # cover -- for a chrom-sharded panel this skips the shards for every other
+  # chromosome. Then run the per-study containment checks (deferred until the
+  # sketch exists).
+  ldSketch <- .materializeLdSketch(ldSketchSpec, .entriesChroms(entries))
+  for (i in seq_len(nrow(df))) {
+    .checkLdContainment(ldSketch, entries[[i]], minLdOverlapWarn,
+                        paste0("GwasSumStats[study=", df$study[[i]], "]"))
+  }
+
   # Trim a genome-wide LD sketch to the summary stats' per-chromosome position
   # span so the object doesn't carry a full-genome snpInfo.
   ldSketch <- .subsetSketchToRange(ldSketch, entries)
@@ -945,12 +982,13 @@ loadGwasSumStatsFromManifest <- function(manifest, genome = NULL,
   do.call(GwasSumStats, args)
 }
 
-# Build the entry list + tuple vectors for a QtlSumStats manifest. `allowNoN`
-# is an optional per-row logical vector: when TRUE for a row, its sumstats file
-# need not carry a per-variant N (a study-level nSample scalar fills it later in
-# summaryStatsQc). NULL (default) requires a per-variant N on every row.
-.loadQtlSumStatsEntries <- function(df, base, region, ldSketch,
-                                    minLdOverlapWarn, columnMapping,
+# Build the entry list for a QtlSumStats manifest. `allowNoN` is an optional
+# per-row logical vector: when TRUE for a row, its sumstats file need not carry
+# a per-variant N (a study-level nSample scalar fills it later in
+# summaryStatsQc). NULL (default) requires a per-variant N on every row. The LD
+# containment check is deferred to the caller so the sketch can first be read
+# restricted to the chromosomes these entries cover.
+.loadQtlSumStatsEntries <- function(df, base, region, columnMapping,
                                     sampleSelect, formatMapping,
                                     allowNoN = NULL) {
   lapply(seq_len(nrow(df)), function(i) {
@@ -963,13 +1001,9 @@ loadGwasSumStatsFromManifest <- function(manifest, genome = NULL,
     } else {
       columnMapping
     }
-    gr <- .loadSumStatsEntry(.resolveRel(as.character(df$sumStatsPath[[i]]), base),
-                             region, mapping, sampleSelect, formatMapping, label,
-                             allowNoN = !is.null(allowNoN) && isTRUE(allowNoN[[i]]))
-    if (!is.null(ldSketch)) {
-      .checkLdContainment(ldSketch, gr, minLdOverlapWarn, label)
-    }
-    gr
+    .loadSumStatsEntry(.resolveRel(as.character(df$sumStatsPath[[i]]), base),
+                       region, mapping, sampleSelect, formatMapping, label,
+                       allowNoN = !is.null(allowNoN) && isTRUE(allowNoN[[i]]))
   })
 }
 
@@ -1008,7 +1042,7 @@ loadQtlSumStatsFromManifest <- function(manifest, genome = NULL,
                                         "sumStatsPath"),
                            label = "QtlSumStats")
   genome <- .reconcileScalar(df$genome, genome, "genome")
-  ldSketch <- .resolveLdSketchInput(df, ldSketch, base)
+  ldSketchSpec <- .resolveLdSketchInput(df, ldSketch, base)
 
   # Tuple-level total-N scalar (from the manifest). When a row carries a usable
   # nSample the sumstats file need not supply a per-variant N: summaryStatsQc
@@ -1016,10 +1050,21 @@ loadQtlSumStatsFromManifest <- function(manifest, genome = NULL,
   nSampleCol <- if ("nSample" %in% names(df)) as.numeric(df$nSample) else NULL
   allowNoN   <- if (!is.null(nSampleCol)) is.finite(nSampleCol) else NULL
 
-  entries <- .loadQtlSumStatsEntries(df, base, region, ldSketch,
-                                     minLdOverlapWarn, columnMapping,
+  entries <- .loadQtlSumStatsEntries(df, base, region, columnMapping,
                                      sampleSelect, formatMapping,
                                      allowNoN = allowNoN)
+
+  # Materialise the LD sketch reading only the chromosomes the summary stats
+  # cover (skips other shards of a chrom-sharded panel), then check containment.
+  ldSketch <- .materializeLdSketch(ldSketchSpec, .entriesChroms(entries))
+  if (!is.null(ldSketch)) {
+    for (i in seq_len(nrow(df))) {
+      .checkLdContainment(ldSketch, entries[[i]], minLdOverlapWarn,
+                          paste0("QtlSumStats[", df$study[[i]], "/",
+                                 df$context[[i]], "/", df$trait[[i]], "]"))
+    }
+  }
+
   # Trim a genome-wide LD sketch to the summary stats' per-chromosome position
   # span so the object doesn't carry a full-genome snpInfo.
   ldSketch <- .subsetSketchToRange(ldSketch, entries)
diff --git a/R/sumstatsQc.R b/R/sumstatsQc.R
index c379042d..7c9b180b 100644
--- a/R/sumstatsQc.R
+++ b/R/sumstatsQc.R
@@ -2614,19 +2614,105 @@ krigingOutlierQc <- function(zScore, R, n, variantIds = NULL,
        diagnostics = diagnostics)
 }
 
-# Per-entry SER-based pip-screen (skip if no signal above the cutoff).
-.applyPipScreen <- function(df, n, cutoff) {
-  if (cutoff <= 0) return(list(df = df, skipped = FALSE))
-  effectiveCutoff <- if (cutoff < 0) 3 / nrow(df) else cutoff
-  pip <- susieR::susie_ser(z = df$Z, n = n, coverage = NULL)$pip
-  if (!any(pip > effectiveCutoff)) {
+# -----------------------------------------------------------------------------
+# Signal screen: skip an entry/block with no strong signal by a chosen metric.
+# -----------------------------------------------------------------------------
+# The screen is driven by ONE metric at a time (enforced by .resolveScreenMetric):
+#   pip   : max single-effect PIP                (susie_ser $pip); cutoff<0 => 3/nVar
+#   absZ  : max |Z|                              (no model fit)
+#   logBf : max per-variant single-effect logBF  (susie_ser $lbf_variable)
+#   bf    : same evidence on the raw BF scale    (compare maxlogBF > log(cutoff))
+# A "screen spec" flowing through the pipelines is EITHER a legacy PIP cutoff
+# (numeric scalar, 0 = off -- the historical `pipCutoffToSkip`) OR a resolved
+# screen object `list(metric, cutoff)`. .asScreen() canonicalizes either into
+# `list(metric, cutoff)` or NULL (no screen); the pipeline channels stay
+# untyped so only the screen primitives need to interpret the spec.
+
+# Canonicalize a screen spec into list(metric, cutoff) or NULL (no screen).
+.asScreen <- function(spec) {
+  if (is.null(spec)) return(NULL)
+  if (is.list(spec)) {                                   # already a screen object
+    if (is.null(spec$metric) || is.null(spec$cutoff) ||
+        is.na(spec$cutoff) || spec$cutoff == 0) return(NULL)
+    return(spec)
+  }
+  # Legacy numeric: a PIP cutoff. Only a non-zero scalar activates it (a
+  # non-scalar / NA / 0 means no screen), matching the historical behaviour.
+  if (length(spec) != 1L || is.na(spec) || spec == 0) return(NULL)
+  list(metric = "pip", cutoff = as.numeric(spec))
+}
+
+# Turn the four public cutoff arguments into a single screen object (or NULL).
+# Enforces one-metric-at-a-time and rejects meaningless negative cutoffs. The
+# pip metric keeps its `< 0 => 3 / nVariants` adaptive convention (resolved at
+# screen time); absZ / bf must be > 0; logBf may be any non-zero value.
+.resolveScreenMetric <- function(pipCutoffToSkip = 0, absZCutoffToSkip = 0,
+                                 bfCutoffToSkip = 0, logBfCutoffToSkip = 0) {
+  scalar <- function(x) if (is.null(x) || length(x) != 1L || is.na(x)) 0 else as.numeric(x)
+  cuts <- c(pip = scalar(pipCutoffToSkip), absZ = scalar(absZCutoffToSkip),
+            bf = scalar(bfCutoffToSkip), logBf = scalar(logBfCutoffToSkip))
+  on <- cuts[cuts != 0]
+  if (length(on) == 0L) return(NULL)
+  if (length(on) > 1L)
+    stop("Only one signal screen may be enabled at a time, but these are ",
+         "non-zero: ", paste(sprintf("%s=%g", names(on), on), collapse = ", "),
+         ". Set all but one of pipCutoffToSkip / absZCutoffToSkip / ",
+         "bfCutoffToSkip / logBfCutoffToSkip to 0.")
+  metric <- names(on); cutoff <- unname(on[[1L]])
+  if (metric == "absZ" && cutoff < 0)
+    stop("absZCutoffToSkip must be > 0 (it screens on max|Z|).")
+  if (metric == "bf" && cutoff < 0)
+    stop("bfCutoffToSkip must be > 0 (Bayes factors are positive).")
+  list(metric = metric, cutoff = cutoff)
+}
+
+# Decide whether the z-scores of one entry clear the chosen screen. Returns
+# list(ok = logical, reason = character). susie_ser is fit at most once, and
+# only for the metrics that need it (absZ stays model-free).
+.entryScreenPass <- function(z, n, nVar, scr) {
+  metric <- scr$metric; cutoff <- scr$cutoff
+  if (metric == "absZ") {
+    m <- suppressWarnings(max(abs(as.numeric(z)), na.rm = TRUE))
+    return(list(ok = is.finite(m) && m > cutoff,
+                reason = sprintf("no variant with |Z| above %g (max |Z| = %g)",
+                                 cutoff, m)))
+  }
+  ser <- susieR::susie_ser(z = z, n = n, coverage = NULL)
+  if (metric == "pip") {
+    eff <- if (cutoff < 0) 3 / nVar else cutoff
+    return(list(ok = any(ser$pip > eff),
+                reason = sprintf("no signals above PIP threshold %g", eff)))
+  }
+  maxLbf <- suppressWarnings(max(as.numeric(ser$lbf_variable), na.rm = TRUE))
+  if (metric == "logBf")
+    return(list(ok = is.finite(maxLbf) && maxLbf > cutoff,
+                reason = sprintf("no variant with logBF above %g (max logBF = %g)",
+                                 cutoff, maxLbf)))
+  # metric == "bf": compare in log space to avoid overflow of exp(maxLbf).
+  list(ok = is.finite(maxLbf) && maxLbf > log(cutoff),
+       reason = sprintf("no variant with BF above %g (max BF = %g)",
+                        cutoff, exp(maxLbf)))
+}
+
+# Per-entry signal screen. `screen` is a screen spec (see .asScreen). Skips
+# (empties) the entry when the chosen metric shows no signal above its cutoff.
+.applyEntryScreen <- function(df, n, screen) {
+  scr <- .asScreen(screen)
+  if (is.null(scr)) return(list(df = df, skipped = FALSE))
+  res <- .entryScreenPass(df$Z, n = n, nVar = nrow(df), scr = scr)
+  if (!res$ok) {
     return(list(df = df[FALSE, , drop = FALSE], skipped = TRUE,
-                reason = sprintf("no signals above PIP threshold %g",
-                                 effectiveCutoff)))
+                reason = res$reason))
   }
   list(df = df, skipped = FALSE)
 }
 
+# Back-compat thin wrapper for the original PIP-only screen (a bare numeric
+# cutoff, 0 = off). Delegates to the generalized screen via .asScreen.
+.applyPipScreen <- function(df, n, cutoff) {
+  .applyEntryScreen(df, n = n, screen = cutoff)
+}
+
 # Prefix QC-track log lines with the entry label `lbl` (as `[lbl] ...`), or emit
 # them bare when `lbl` is NA.
 # @noRd
@@ -2839,12 +2925,13 @@ krigingOutlierQc <- function(zScore, R, n, variantIds = NULL,
          " variant(s).")
   }
 
-  # 5. Optional PIP screen (after harmonization, on panel-aligned variants).
-  if (opts$pipCutoffToSkip != 0) {
-    pip <- .applyPipScreen(df, n = opts$nForPip, cutoff = opts$pipCutoffToSkip)
-    df <- pip$df
-    entryAudit$pipScreenSkipped <- isTRUE(pip$skipped)
-    if (isTRUE(pip$skipped)) entryAudit$pipScreenReason <- pip$reason
+  # 5. Optional signal screen (after harmonization, on panel-aligned variants).
+  #    One metric at a time: PIP / max|Z| / max BF / max logBF (opts$screen).
+  if (!is.null(opts$screen)) {
+    scr <- .applyEntryScreen(df, n = opts$nForPip, screen = opts$screen)
+    df <- scr$df
+    entryAudit$pipScreenSkipped <- isTRUE(scr$skipped)
+    if (isTRUE(scr$skipped)) entryAudit$pipScreenReason <- scr$reason
   }
 
   # 6. Optional kriging allele-flip QC. Uses susieR's allele-switch rule
@@ -3140,6 +3227,20 @@ krigingOutlierQc <- function(zScore, R, n, variantIds = NULL,
 #'   LD-independent single-effect SER screen and skip the entry if no
 #'   PIP exceeds the cutoff. \code{< 0} resolves to \code{3 / nVariants}.
 #'   Default 0 (no screen).
+#' @param absZCutoffToSkip Numeric (length 1). Alternative signal screen:
+#'   skip the entry when \code{max(abs(Z))} does not exceed the cutoff. No
+#'   model fit. Default 0 (off).
+#' @param bfCutoffToSkip Numeric (length 1). Alternative signal screen: skip
+#'   the entry when the largest per-variant single-effect Bayes factor (from
+#'   the same \code{susie_ser} fit as the PIP screen) does not exceed the
+#'   cutoff. Compared in log space (\code{maxlogBF > log(cutoff)}). Must be
+#'   \code{> 0}. Default 0 (off).
+#' @param logBfCutoffToSkip Numeric (length 1). As \code{bfCutoffToSkip} but
+#'   the cutoff is on the log Bayes factor scale (\code{maxlogBF > cutoff}).
+#'   Default 0 (off).
+#'   Exactly one of \code{pipCutoffToSkip} / \code{absZCutoffToSkip} /
+#'   \code{bfCutoffToSkip} / \code{logBfCutoffToSkip} may be non-zero (one
+#'   screening metric at a time); enabling more than one is an error.
 #' @param zMismatchQc One of \code{"none"} (default), \code{"slalom"},
 #'   \code{"dentist"}.
 #' @param alleleFlipKriging Logical (length 1). Opt-in kriging
@@ -3205,6 +3306,9 @@ summaryStatsQc <- function(sumstats,
                            keepVariants           = NULL,
                            skipRegion             = NULL,
                            pipCutoffToSkip        = 0,
+                           absZCutoffToSkip       = 0,
+                           bfCutoffToSkip         = 0,
+                           logBfCutoffToSkip      = 0,
                            zMismatchQc            = c("none", "slalom",
                                                      "dentist"),
                            alleleFlipKriging      = FALSE,
@@ -3250,7 +3354,10 @@ summaryStatsQc <- function(sumstats,
     nCutoff                = nCutoff,
     keepVariants           = as.character(keepVariants),
     skipRegion             = skipRegion,
-    pipCutoffToSkip        = pipCutoffToSkip,
+    screen                 = .resolveScreenMetric(pipCutoffToSkip,
+                                                  absZCutoffToSkip,
+                                                  bfCutoffToSkip,
+                                                  logBfCutoffToSkip),
     zMismatchQc            = zMismatchQc,
     alleleFlipKriging      = alleleFlipKriging,
     effectiveN             = effectiveN,
@@ -3312,6 +3419,10 @@ summaryStatsQc <- function(sumstats,
       mafCutoff             = mafCutoff,
       infoCutoff            = infoCutoff,
       nCutoff               = nCutoff,
+      pipCutoffToSkip       = pipCutoffToSkip,
+      absZCutoffToSkip      = absZCutoffToSkip,
+      bfCutoffToSkip        = bfCutoffToSkip,
+      logBfCutoffToSkip     = logBfCutoffToSkip,
       zMismatchQc           = zMismatchQc,
       alleleFlipKriging     = alleleFlipKriging,
       effectiveN            = effectiveN,
diff --git a/man/GenotypeHandle.Rd b/man/GenotypeHandle.Rd
index c6fe0d8f..30f47cab 100644
--- a/man/GenotypeHandle.Rd
+++ b/man/GenotypeHandle.Rd
@@ -17,6 +17,7 @@ GenotypeHandle(
   ldMeta = NULL,
   region = NULL,
   genoMeta = NULL,
+  chroms = NULL,
   ...
 )
 }
@@ -51,6 +52,17 @@ supported here — use \code{\link{loadLdMatrix}} for that case.}
 (names = chromosomes, values = payload paths/prefixes). Optionally pass
 \code{format} via \code{...} to force a single backend for every shard.}
 
+\item{chroms}{Optional character vector of chromosomes. With \code{genoMeta}
+(a sharded, one-file-per-chromosome panel) only the shards whose
+chromosome is listed are read, skipping the rest -- an I/O optimisation
+when the panel is genome-wide but only a few chromosomes are needed.
+Chromosome labels are compared canonically (\code{"chr1"}/\code{"1"} match,
+\code{23}/\code{X} etc.). If none of the requested chromosomes are present
+in the panel, every shard is read (so the caller's own absence check can
+report the mismatch). Only meaningful with \code{genoMeta}; supplying it
+with any other source is an error (a single-file panel has no per-chromosome
+shards to skip).}
+
 \item{...}{Additional arguments forwarded to the format-specific reader.}
 }
 \value{
diff --git a/man/colocboostPipeline.Rd b/man/colocboostPipeline.Rd
index b8ede09a..51ae8374 100644
--- a/man/colocboostPipeline.Rd
+++ b/man/colocboostPipeline.Rd
@@ -23,6 +23,9 @@ colocboostPipeline(qtlData, gwasSumStats = NULL, ...)
   separateGwas = FALSE,
   samples = NULL,
   pipCutoffToSkip = 0,
+  absZCutoffToSkip = 0,
+  bfCutoffToSkip = 0,
+  logBfCutoffToSkip = 0,
   alleleFlip = TRUE,
   ...
 )
@@ -55,6 +58,9 @@ colocboostPipeline(qtlData, gwasSumStats = NULL, ...)
   separateGwas = FALSE,
   samples = NULL,
   pipCutoffToSkip = 0,
+  absZCutoffToSkip = 0,
+  bfCutoffToSkip = 0,
+  logBfCutoffToSkip = 0,
   alleleFlip = TRUE,
   ...
 )
@@ -106,6 +112,15 @@ skipped. \code{0} (default) disables it; a negative value uses
 \code{3 / n_variants}. (Summary-statistic skipping is handled upstream by
 \code{\link{summaryStatsQc}}'s own \code{pipCutoffToSkip}.)}
 
+\item{absZCutoffToSkip, bfCutoffToSkip, logBfCutoffToSkip}{Alternative
+individual-level pre-filter metrics used in place of
+\code{pipCutoffToSkip}: drop an outcome unless its maximum marginal
+\code{|z|} (\code{absZCutoffToSkip}), or its maximum per-variant
+single-effect Bayes factor (\code{bfCutoffToSkip}) / log Bayes factor
+(\code{logBfCutoffToSkip}) from the \code{L = 1} fit, exceeds the cutoff.
+Scalars, each defaulting to 0 (off). Exactly one screening metric may be
+enabled: setting any of these requires \code{pipCutoffToSkip = 0}.}
+
 \item{alleleFlip}{Logical, default \code{TRUE}. When TRUE, harmonize variants
 across the individual X, sumstats, and LD by (chrom, pos) with ref/alt
 swaps recognized (flipping z / residualized dosage / LD to a shared
diff --git a/man/fineMappingPipeline.Rd b/man/fineMappingPipeline.Rd
index a65185f7..02db1a6c 100644
--- a/man/fineMappingPipeline.Rd
+++ b/man/fineMappingPipeline.Rd
@@ -40,6 +40,9 @@ fineMappingPipeline(data, ...)
   cvThreads = 1,
   samplePartition = NULL,
   pipCutoffToSkip = 0,
+  absZCutoffToSkip = 0,
+  bfCutoffToSkip = 0,
+  logBfCutoffToSkip = 0,
   usePCA = FALSE,
   nPCs = 10L,
   seed = NULL,
@@ -80,6 +83,9 @@ fineMappingPipeline(data, ...)
   cvThreads = 1,
   samplePartition = NULL,
   pipCutoffToSkip = 0,
+  absZCutoffToSkip = 0,
+  bfCutoffToSkip = 0,
+  logBfCutoffToSkip = 0,
   seed = NULL,
   naAction = c("drop", "impute"),
   verbose = 1,
@@ -261,6 +267,15 @@ summary-statistics analog lives in \code{summaryStatsQc()}. \code{0}
 (default) disables the screen; a negative value uses the adaptive
 \code{3 / nVariants} threshold.}
 
+\item{absZCutoffToSkip, bfCutoffToSkip, logBfCutoffToSkip}{Numeric (length 1).
+Alternative individual-level pre-screen metrics, in place of
+\code{pipCutoffToSkip}: skip a block unless its maximum marginal
+\code{|z|} (\code{absZCutoffToSkip}), or its maximum per-variant
+single-effect Bayes factor (\code{bfCutoffToSkip}) / log Bayes factor
+(\code{logBfCutoffToSkip}) from the \code{L = 1} fit, exceeds the cutoff.
+Each defaults to 0 (off). Exactly one of the four \code{*CutoffToSkip}
+arguments may be non-zero (one screening metric at a time).}
+
 \item{usePCA}{Logical (length 1). \code{QtlDataset} only. When
 \code{TRUE} (default \code{FALSE}), each multi-trait context's
 PCA-reduced phenotype is fine-mapped with univariate SuSiE on its
diff --git a/man/summaryStatsQc.Rd b/man/summaryStatsQc.Rd
index baf8a069..0a562106 100644
--- a/man/summaryStatsQc.Rd
+++ b/man/summaryStatsQc.Rd
@@ -14,6 +14,9 @@ summaryStatsQc(
   keepVariants = NULL,
   skipRegion = NULL,
   pipCutoffToSkip = 0,
+  absZCutoffToSkip = 0,
+  bfCutoffToSkip = 0,
+  logBfCutoffToSkip = 0,
   zMismatchQc = c("none", "slalom", "dentist"),
   alleleFlipKriging = FALSE,
   effectiveN = TRUE,
@@ -64,6 +67,23 @@ LD-independent single-effect SER screen and skip the entry if no
 PIP exceeds the cutoff. \code{< 0} resolves to \code{3 / nVariants}.
 Default 0 (no screen).}
 
+\item{absZCutoffToSkip}{Numeric (length 1). Alternative signal screen:
+skip the entry when \code{max(abs(Z))} does not exceed the cutoff. No
+model fit. Default 0 (off).}
+
+\item{bfCutoffToSkip}{Numeric (length 1). Alternative signal screen: skip
+the entry when the largest per-variant single-effect Bayes factor (from
+the same \code{susie_ser} fit as the PIP screen) does not exceed the
+cutoff. Compared in log space (\code{maxlogBF > log(cutoff)}). Must be
+\code{> 0}. Default 0 (off).}
+
+\item{logBfCutoffToSkip}{Numeric (length 1). As \code{bfCutoffToSkip} but
+the cutoff is on the log Bayes factor scale (\code{maxlogBF > cutoff}).
+Default 0 (off).
+Exactly one of \code{pipCutoffToSkip} / \code{absZCutoffToSkip} /
+\code{bfCutoffToSkip} / \code{logBfCutoffToSkip} may be non-zero (one
+screening metric at a time); enabling more than one is an error.}
+
 \item{zMismatchQc}{One of \code{"none"} (default), \code{"slalom"},
 \code{"dentist"}.}
 
diff --git a/tests/testthat/test_colocboostPipeline.R b/tests/testthat/test_colocboostPipeline.R
index 40408dde..63c31123 100644
--- a/tests/testthat/test_colocboostPipeline.R
+++ b/tests/testthat/test_colocboostPipeline.R
@@ -220,6 +220,32 @@ test_that(".cbBuildLdArgs: empty list returns empty list", {
   expect_equal(pecotmr:::.cbBuildLdArgs(list()), list())
 })
 
+test_that(".cbScreenSpec enforces one metric and threads the right spec", {
+  expect_error(pecotmr:::.cbScreenSpec(0.5, 5, 0, 0), "one signal screen")
+  expect_equal(pecotmr:::.cbScreenSpec(0, 5, 0, 0), list(metric = "absZ", cutoff = 5))
+  expect_equal(pecotmr:::.cbScreenSpec(0, 0, 100, 0), list(metric = "bf", cutoff = 100))
+  expect_equal(pecotmr:::.cbScreenSpec(0.3, 0, 0, 0), 0.3)          # legacy pip scalar
+  expect_equal(pecotmr:::.cbScreenSpec(c(brain = 0.3), 0, 0, 0),
+               c(brain = 0.3))                                       # context-named pass-through
+})
+
+test_that(".cbResolveCutoff passes a screen object through uniformly", {
+  sc <- list(metric = "absZ", cutoff = 5)
+  expect_identical(pecotmr:::.cbResolveCutoff(sc, "brain"), sc)
+  expect_identical(pecotmr:::.cbResolveCutoff(sc, "blood"), sc)
+  expect_equal(pecotmr:::.cbResolveCutoff(c(brain = 0.3), "brain"), 0.3)
+  expect_equal(pecotmr:::.cbResolveCutoff(c(brain = 0.3), "blood"), 0)  # unlisted ctx
+  expect_equal(pecotmr:::.cbResolveCutoff(0.5, "any"), 0.5)
+})
+
+test_that("colocboostPipeline(QtlDataset): enabling two screen metrics errors", {
+  qd <- .cbp_makeQtlDataset(contexts = "brain", traits = "ENSG_A")
+  # .cbScreenSpec runs before the bundle/engine, so this fails fast.
+  expect_error(
+    colocboostPipeline(qd, pipCutoffToSkip = 0.5, bfCutoffToSkip = 100),
+    "one signal screen")
+})
+
 test_that(".cbRequireSumStatsQc: un-QCd input errors", {
   ss <- .cbp_makeQtlSumStats(qc = FALSE)
   expect_error(
@@ -564,7 +590,7 @@ test_that(".cbPipSkipOutcomes: an outcome with < 2 observations is skipped", {
   set.seed(4)
   X <- matrix(rnorm(60), 30, 2, dimnames = list(paste0("s", 1:30), c("v1", "v2")))
   Y <- cbind(a = c(1, rep(NA, 29)), b = rnorm(30))          # col a: 1 obs (< 2)
-  res <- pecotmr:::.cbPipSkipOutcomes(X, Y, cutoff = 0.5)
+  res <- pecotmr:::.cbPipSkipOutcomes(X, Y, 0.5)
   expect_true(is.null(res) || is.matrix(res))               # col a -> next (180)
 })
 
diff --git a/tests/testthat/test_ctwasPipeline.R b/tests/testthat/test_ctwasPipeline.R
index 4d71a90b..6ebacacb 100644
--- a/tests/testthat/test_ctwasPipeline.R
+++ b/tests/testthat/test_ctwasPipeline.R
@@ -1267,7 +1267,8 @@ test_that("estCtwasParam: fallbackToPrefit recovers from accurate-EM NaN diverge
   # produce a stub prefit result. Verify estCtwasParam catches the
   # NaN error AND that the returned param is the prefit estimate.
   local_mocked_bindings(
-    assemble_region_data = function(...) list(block1 = list(stub = TRUE)),
+    assemble_region_data = function(...) list(block1 = list(
+      gid = "t1", sid = c("s1", "s2"), stub = TRUE)),
     get_boundary_genes   = function(...) data.frame(id = "t1", n_regions = 2L),
     compute_gene_z = function(...) data.frame(id = "t1", z = 1.0),
     est_param = function(...) stop("Estimated group_prior_var contains NAs!"),
@@ -1295,6 +1296,61 @@ test_that("estCtwasParam: fallbackToPrefit recovers from accurate-EM NaN diverge
   expect_equal(unname(est$param$group_prior_var), c(4.0, 5.0))
 })
 
+test_that("estCtwasParam fallback drops degenerate regions before fit_EM", {
+  # Regression for the ctwas >= 0.6.0 breakage: the prefit fallback used to hand
+  # ALL regions to ctwas::fit_EM, so a degenerate region (empty gid/sid, whose
+  # `sid` ctwas::extract_region_data now requires) crashed with
+  # "regiondata$sid ... target is NULL". The fallback must mirror est_param's
+  # min_var / min_gene skip and fit only the qualifying regions.
+  skip_if_not_installed("ctwas")
+  inp  <- .ctp_makeMultiBlockInputs()
+  seen <- NULL
+  local_mocked_bindings(
+    assemble_region_data = function(...) list(
+      good       = list(gid = "t1", sid = c("s1", "s2")),  # 1 gene + 2 SNPs -> kept
+      degenerate = list(gid = character(0), sid = NULL)),  # no variables    -> dropped
+    get_boundary_genes = function(...) data.frame(id = "t1", n_regions = 2L),
+    compute_gene_z     = function(...) data.frame(id = "t1", z = 1.0),
+    est_param          = function(...) stop("No regions selected!"),
+    fit_EM = function(region_data, ...) {
+      seen <<- names(region_data)
+      list(group_prior     = c(g = 0.05, SNP = 1e-4),
+           group_prior_var = c(g = 4.0,  SNP = 5.0),
+           group_size      = c(g = 1,    SNP = 100))
+    },
+    .package = "ctwas")
+  local_mocked_bindings(extractBlockGenotypes = .ctp_mockExtractor(),
+                        .package = "pecotmr")
+  est <- estCtwasParam(
+    assembleCtwasInputs(inp$gwasSumStats, inp$twasWeights),
+    fallbackToPrefit = TRUE)
+  # only the qualifying region reached fit_EM; the degenerate region was filtered
+  expect_equal(seen, "good")
+  # ...but every region is still accounted for in the returned p_single_effect
+  expect_setequal(est$param$p_single_effect$region_id, c("good", "degenerate"))
+})
+
+test_that("(real ctwas) prefit fallback skips a degenerate region fit_EM would reject", {
+  # No-mock guard for the ctwas >= 0.6.0 contract. Runs the REAL ctwas::fit_EM
+  # (via .ctwasFitPrefitEm) on a genuine assemble_region_data fixture with one
+  # valid region (1 gene, 108 SNPs) and one degenerate region (0 genes/SNPs,
+  # unset `sid`). Handing the degenerate region to fit_EM crashes ctwas >= 0.6.0
+  # in extract_region_data ("regiondata$sid ... target is NULL"); the fallback
+  # must filter it. Unlike the mocked tests above, this exercises the real engine,
+  # so it would catch a FUTURE ctwas contract change (which the mocks cannot).
+  skip_if_not_installed("ctwas")
+  region_data <- readRDS(test_path("test_data", "ctwas_region_data_degenerate.rds"))
+  expect_length(region_data, 2L)
+  res <- .ctwasFitPrefitEm(
+    region_data, niterPrefit = 3L,
+    groupPriorVarStructure = "shared_all", thin = 1, ncore = 1L)
+  # the prefit EM ran on the valid region only and returns finite real group priors
+  expect_true("SNP" %in% names(res$group_prior))
+  expect_true(all(is.finite(res$group_prior)))
+  # every region (valid + degenerate) is still listed in p_single_effect
+  expect_setequal(res$p_single_effect$region_id, names(region_data))
+})
+
 test_that("estCtwasParam / screenCtwasRegions / finemapCtwasRegions can be called independently", {
   skip_if_not_installed("ctwas")
   inp <- .ctp_makeMultiBlockInputs()
diff --git a/tests/testthat/test_data/ctwas_region_data_degenerate.rds b/tests/testthat/test_data/ctwas_region_data_degenerate.rds
new file mode 100644
index 00000000..bd5bd2c6
Binary files /dev/null and b/tests/testthat/test_data/ctwas_region_data_degenerate.rds differ
diff --git a/tests/testthat/test_fineMappingPipeline.R b/tests/testthat/test_fineMappingPipeline.R
index 2504c735..bead0c56 100644
--- a/tests/testthat/test_fineMappingPipeline.R
+++ b/tests/testthat/test_fineMappingPipeline.R
@@ -846,6 +846,38 @@ test_that(".fmBuildMvsusiePriorCv: mode C reuses full-fit w0/V with per-fold U",
   expect_equal(captured[[2]]$mixture_prior$matrices$Z, diag(2) * 3)  # fold 2's U
 })
 
+test_that(".fmSerScreen supports absZ / bf / logBf metrics and the legacy pip scalar", {
+  skip_if_not_installed("susieR")
+  set.seed(11)
+  n <- 200L; p <- 6L
+  X <- matrix(stats::rnorm(n * p), n, p)
+  yStrong <- X[, 2] * 0.6 + stats::rnorm(n)   # column 2 strongly associated
+  yNull   <- stats::rnorm(n)                  # no association
+
+  # absZ: max marginal |z| (no susie fit).
+  expect_true(pecotmr:::.fmSerScreen(X, yStrong, list(metric = "absZ", cutoff = 3)))
+  expect_false(pecotmr:::.fmSerScreen(X, yNull, list(metric = "absZ", cutoff = 3)))
+  # bf / logBf from the L = 1 susie lbf_variable.
+  expect_true(pecotmr:::.fmSerScreen(X, yStrong, list(metric = "logBf", cutoff = 2)))
+  expect_false(pecotmr:::.fmSerScreen(X, yNull,  list(metric = "logBf", cutoff = 5)))
+  expect_true(pecotmr:::.fmSerScreen(X, yStrong, list(metric = "bf", cutoff = 10)))
+  # Legacy scalar spec still screens on PIP; 0 disables (always keep).
+  expect_true(pecotmr:::.fmSerScreen(X, yStrong, 0.5))
+  expect_true(pecotmr:::.fmSerScreen(X, yNull, 0))
+  # Too few samples -> advisory fallback (keep by default).
+  expect_true(pecotmr:::.fmSerScreen(X[1, , drop = FALSE], yStrong[1],
+                                     list(metric = "absZ", cutoff = 3)))
+})
+
+test_that("fineMappingPipeline(QtlDataset): enabling two screen metrics errors", {
+  qd <- .fmp_makeQtlDataset(contexts = "brain", traits = "ENSG_A")
+  # The resolver runs before any fitting, so this fails fast on the public call.
+  expect_error(
+    fineMappingPipeline(qd, methods = "susie", cisWindow = 1000L,
+                        pipCutoffToSkip = 0.5, absZCutoffToSkip = 5),
+    "one signal screen")
+})
+
 test_that("fineMappingPipeline(QtlDataset): pipCutoffToSkip skips no-signal univariate traits", {
   qd <- .fmp_makeQtlDataset(contexts = "brain", traits = c("ENSG_A", "ENSG_B"))
   # Stateful screen: reject the first block (ENSG_A), keep the rest (ENSG_B).
@@ -2659,7 +2691,7 @@ test_that(".fmSerScreen / .fmScreenActive / .fmSerScreenColumns", {
   set.seed(2)
   X <- matrix(rnorm(40), 20, 2, dimnames = list(paste0("s", 1:20), c("v1", "v2")))
   y <- rnorm(20)
-  expect_true(pecotmr:::.fmSerScreen(X, y, cutoff = 0))              # disabled
+  expect_true(pecotmr:::.fmSerScreen(X, y, 0))                       # disabled
   expect_true(pecotmr:::.fmSerScreen(X, c(1, rep(NA, 19)), 0.5))     # < 2 obs (880)
   expect_type(pecotmr:::.fmSerScreen(X, y, 0.5), "logical")          # real susie fit
   local_mocked_bindings(susie = function(...) stop("boom"), .package = "susieR")
diff --git a/tests/testthat/test_genotypeHandle.R b/tests/testthat/test_genotypeHandle.R
index 1482f048..3a56852f 100644
--- a/tests/testthat/test_genotypeHandle.R
+++ b/tests/testthat/test_genotypeHandle.R
@@ -428,6 +428,62 @@ test_that("genoMeta sharded handle show() reports the layout", {
   expect_match(out, "per-chromosome files")
 })
 
+# ===========================================================================
+# genoMeta `chroms`: read only the requested per-chromosome shards (I/O win).
+# ===========================================================================
+test_that("genoMeta chroms reads only the requested shard", {
+  skip_if_not_installed("snpStats")
+  meta <- c("21" = file.path(test_data_dir, "test_variants"),
+            "22" = file.path(test_data_dir, "test_variants_chr22"))
+  full   <- GenotypeHandle(genoMeta = meta)
+  only21 <- GenotypeHandle(genoMeta = meta, chroms = "21")
+  expect_equal(names(only21@chromPaths), "21")
+  expect_equal(nrow(only21@snpInfo), 349L)
+  # chr21 is the first shard, so the kept rows/fileIdx match the full handle's.
+  expect_equal(only21@snpInfo$SNP,     full@snpInfo$SNP[1:349])
+  expect_equal(only21@snpInfo$fileIdx, full@snpInfo$fileIdx[1:349])
+})
+
+test_that("genoMeta chroms canonicalises chromosome labels", {
+  skip_if_not_installed("snpStats")
+  # "chr21" canonicalises to the panel's declared "21" shard.
+  h <- GenotypeHandle(
+    genoMeta = c("21" = file.path(test_data_dir, "test_variants"),
+                 "22" = file.path(test_data_dir, "test_variants_chr22")),
+    chroms = "chr21")
+  expect_equal(names(h@chromPaths), "21")
+})
+
+test_that("genoMeta chroms falls back to all shards when none match", {
+  skip_if_not_installed("snpStats")
+  # chr9 is absent from the panel: rather than build an empty handle, read all
+  # so the caller's own absence check can report the mismatch.
+  h <- GenotypeHandle(
+    genoMeta = c("21" = file.path(test_data_dir, "test_variants"),
+                 "22" = file.path(test_data_dir, "test_variants_chr22")),
+    chroms = "9")
+  expect_equal(sort(names(h@chromPaths)), c("21", "22"))
+})
+
+test_that("genoMeta chroms skips a shard whose file does not exist", {
+  skip_if_not_installed("snpStats")
+  # A bogus chr21 payload proves the skip: restricting to chr22 must never open
+  # it, while requesting chr21 too fails on the missing files.
+  meta <- c("22" = file.path(test_data_dir, "test_variants_chr22"),
+            "21" = "/no/such/chr21/prefix")
+  h <- GenotypeHandle(genoMeta = meta, chroms = "22")
+  expect_equal(names(h@chromPaths), "22")
+  expect_error(GenotypeHandle(genoMeta = meta, chroms = c("22", "21")))
+})
+
+test_that("chroms is rejected for a non-genoMeta source", {
+  skip_if_not_installed("snpStats")
+  expect_error(
+    GenotypeHandle(plink1Prefix = file.path(test_data_dir, "test_variants"),
+                   chroms = "21"),
+    "only supported with .genoMeta")
+})
+
 # ===========================================================================
 # .genotypeHandleFromLdMeta: row-resolution error branches.
 # The real getRegionalLdMeta errors earlier for genuinely-uncovered regions
diff --git a/tests/testthat/test_manifestLoaders.R b/tests/testthat/test_manifestLoaders.R
index 98abd9e3..108e1e8c 100644
--- a/tests/testthat/test_manifestLoaders.R
+++ b/tests/testthat/test_manifestLoaders.R
@@ -545,6 +545,43 @@ test_that(".resolveLdSketch accepts a genoMeta vector, a path, and rejects bad i
   expect_error(pecotmr:::.resolveLdSketch(42L), "must be a GenotypeHandle")
 })
 
+test_that(".entriesChroms collects canonical chromosomes across entries", {
+  a <- pecotmr:::.dfToEntryGranges(.toyGwasDf(3))            # chr22
+  b <- .toyGwasDf(2); b$chrom <- "1"
+  b <- pecotmr:::.dfToEntryGranges(b)
+  expect_setequal(pecotmr:::.entriesChroms(list(a, b)), c("22", "1"))
+  expect_equal(pecotmr:::.entriesChroms(list()), character(0))
+  expect_equal(pecotmr:::.entriesChroms(list(NULL)), character(0))
+})
+
+test_that(".materializeLdSketch passes a handle through and restricts spec shards", {
+  h <- .toyLdSketch()
+  expect_identical(pecotmr:::.materializeLdSketch(h, "22"), h)  # already in memory
+  expect_null(pecotmr:::.materializeLdSketch(NULL, "22"))
+  # A chrom-sharded spec with a bogus chr21 payload: restricting to chr22 must
+  # skip (never open) the chr21 shard; requesting chr21 too fails on it.
+  spec <- c("22" = .toyRefPrefix(), "21" = "/no/such/chr21/prefix")
+  restricted <- pecotmr:::.materializeLdSketch(spec, "22")
+  expect_s4_class(restricted, "GenotypeHandle")
+  expect_equal(names(restricted@chromPaths), "22")
+  expect_error(pecotmr:::.materializeLdSketch(spec, c("22", "21")))
+})
+
+test_that("loadGwasSumStatsFromManifest reads only the sumstats chromosomes' shards", {
+  skip_if_not_installed("snpStats")
+  tmp <- withr::local_tempdir()
+  ssPath <- .writeSumstatsTsv(.toyGwasDf(5), file.path(tmp, "study1.tsv"))
+  manifest <- data.frame(study = "study1", sumStatsPath = ssPath,
+                         stringsAsFactors = FALSE)
+  # Sharded ldSketch: chr22 real (toy_ref), chr21 a bogus path. The sumstats are
+  # all chr22, so the loader must skip the chr21 shard -- opening it would fail.
+  obj <- loadGwasSumStatsFromManifest(
+    manifest, genome = "hg38",
+    ldSketch = c("22" = .toyRefPrefix(), "21" = "/no/such/chr21/prefix"))
+  expect_s4_class(obj, "GwasSumStats")
+  expect_true(methods::validObject(obj))
+})
+
 test_that("ldSketch is resolved from an ldSketchPath column and conflicts error", {
   tmp <- withr::local_tempdir()
   ssPath <- .writeSumstatsTsv(.toyGwasDf(5), file.path(tmp, "s.tsv"))
diff --git a/tests/testthat/test_sumstatsQc.R b/tests/testthat/test_sumstatsQc.R
index 93dc2c4d..4b9d37d2 100644
--- a/tests/testthat/test_sumstatsQc.R
+++ b/tests/testthat/test_sumstatsQc.R
@@ -3601,6 +3601,100 @@ test_that(".applyPipScreen: retains entry when signal clears the cutoff", {
   expect_identical(out$df, df)
 })
 
+# ===========================================================================
+# Signal screen: metric resolver + absZ / bf / logBf metrics
+# ===========================================================================
+
+test_that(".resolveScreenMetric enforces one metric at a time and sane cutoffs", {
+  expect_null(pecotmr:::.resolveScreenMetric())                     # all 0 -> off
+  expect_equal(pecotmr:::.resolveScreenMetric(pipCutoffToSkip = 0.5),
+               list(metric = "pip", cutoff = 0.5))
+  expect_equal(pecotmr:::.resolveScreenMetric(absZCutoffToSkip = 5),
+               list(metric = "absZ", cutoff = 5))
+  expect_equal(pecotmr:::.resolveScreenMetric(bfCutoffToSkip = 100),
+               list(metric = "bf", cutoff = 100))
+  expect_equal(pecotmr:::.resolveScreenMetric(logBfCutoffToSkip = 3),
+               list(metric = "logBf", cutoff = 3))
+  expect_error(pecotmr:::.resolveScreenMetric(pipCutoffToSkip = 0.5,
+                                              absZCutoffToSkip = 5),
+               "one signal screen")
+  expect_error(pecotmr:::.resolveScreenMetric(absZCutoffToSkip = -1), "must be > 0")
+  expect_error(pecotmr:::.resolveScreenMetric(bfCutoffToSkip = -1), "must be > 0")
+})
+
+test_that(".asScreen canonicalizes screen specs", {
+  expect_null(pecotmr:::.asScreen(NULL))
+  expect_null(pecotmr:::.asScreen(0))
+  expect_null(pecotmr:::.asScreen(c(0.1, 0.2)))        # non-scalar numeric -> off
+  expect_equal(pecotmr:::.asScreen(0.9), list(metric = "pip", cutoff = 0.9))
+  sc <- list(metric = "logBf", cutoff = 3)
+  expect_identical(pecotmr:::.asScreen(sc), sc)
+  expect_null(pecotmr:::.asScreen(list(metric = "bf", cutoff = 0)))  # explicit off
+})
+
+test_that(".applyEntryScreen: absZ screen skips / retains on max|Z| (no model fit)", {
+  weak <- data.frame(Z = c(0.2, 0.3, 0.1), stringsAsFactors = FALSE)
+  out <- pecotmr:::.applyEntryScreen(weak, n = 1000,
+                                     screen = list(metric = "absZ", cutoff = 5))
+  expect_true(out$skipped)
+  expect_match(out$reason, "|Z| above 5", fixed = TRUE)
+  expect_equal(nrow(out$df), 0L)
+
+  strong <- data.frame(Z = c(0.2, 6, 0.1), stringsAsFactors = FALSE)
+  out2 <- pecotmr:::.applyEntryScreen(strong, n = 1000,
+                                      screen = list(metric = "absZ", cutoff = 5))
+  expect_false(out2$skipped)
+  expect_identical(out2$df, strong)
+})
+
+test_that(".applyEntryScreen: bf / logBf screens use the SER lbf_variable", {
+  weak   <- data.frame(Z = rep(0.1, 10), stringsAsFactors = FALSE)
+  strong <- data.frame(Z = c(10, 0.1, 0.1, 0.1, 0.1), stringsAsFactors = FALSE)
+  expect_true(pecotmr:::.applyEntryScreen(weak, 1000,
+                list(metric = "logBf", cutoff = 3))$skipped)
+  expect_false(pecotmr:::.applyEntryScreen(strong, 1000,
+                list(metric = "logBf", cutoff = 3))$skipped)
+  expect_true(pecotmr:::.applyEntryScreen(weak, 1000,
+                list(metric = "bf", cutoff = 100))$skipped)
+  expect_false(pecotmr:::.applyEntryScreen(strong, 1000,
+                list(metric = "bf", cutoff = 100))$skipped)
+})
+
+test_that("summaryStatsQc: absZ / bf / logBf screens skip a no-signal entry", {
+  mk <- function() {
+    gr <- .ssQ_makeEntryGr()
+    S4Vectors::mcols(gr)$Z <- rep(0.1, length(gr))
+    GwasSumStats(study = "g1", entry = list(gr), genome = "hg19",
+                 ldSketch = .ssQ_makeHandle())
+  }
+  for (arg in list(list(absZCutoffToSkip = 5),
+                   list(bfCutoffToSkip = 100),
+                   list(logBfCutoffToSkip = 5))) {
+    res <- do.call(summaryStatsQc, c(list(mk()), arg, list(nCutoff = 0)))
+    ea <- getQcInfo(res)$entryAudit[[1L]]
+    expect_true(isTRUE(ea$pipScreenSkipped))
+    expect_equal(length(res$entry[[1L]]), 0L)
+  }
+})
+
+test_that("summaryStatsQc: absZ screen retains an entry with a strong marginal Z", {
+  gr <- .ssQ_makeEntryGr()
+  z <- rep(0.1, length(gr)); z[1] <- 8
+  S4Vectors::mcols(gr)$Z <- z
+  ss <- GwasSumStats(study = "g1", entry = list(gr), genome = "hg19",
+                     ldSketch = .ssQ_makeHandle())
+  res <- summaryStatsQc(ss, absZCutoffToSkip = 5, nCutoff = 0)
+  ea <- getQcInfo(res)$entryAudit[[1L]]
+  expect_false(isTRUE(ea$pipScreenSkipped))
+  expect_gt(length(res$entry[[1L]]), 0L)
+})
+
+test_that("summaryStatsQc: enabling two screens at once errors", {
+  ss <- .ssQ_makeGwasSumStats()
+  expect_error(summaryStatsQc(ss, pipCutoffToSkip = 0.5, absZCutoffToSkip = 5),
+               "one signal screen")
+})
+
 
 context("dentist_qc")
 library(MASS)