ci: install BiasedUrn, which weighted ORA stops without

The Publish Quarto Book run on 2583a25 failed at the Render step with

  The package "BiasedUrn" is required for weighted ORA.

BiasedUrn is in enrichit's Suggests, not its Imports, so pak never installs it
on a fresh runner. enrichment-network.qmd's weighted-ORA chunk calls
ora(weight = w), and enrichit's ora() runs rlang::check_installed('BiasedUrn')
on that path -- it stops rather than degrading to unweighted ORA, so the
chapter cannot render.

Verified locally by hiding the installed package and re-running the chapter's
exact call: the same error appears, and with the package back the weighted
ora() and gsea() both run against the demo GSON.

Fix in three parts, following the rule that a CI dependency list is trimmed by
runtime reachability rather than by counting mentions in the .qmd sources:

- add cran::BiasedUrn to the CRAN install list;
- add it to the verify-loadable step so a future omission is a 10-second
  failure instead of a 21-minute one;
- document the requirement in the chapter, since the dependency is invisible
  in the prose.

Also checked the rest of the dependency surface: the only other packages
named in the chapters that are not installed locally (COSG, createKEGGdb,
KEGG.db, STRINGdb, SeuratData, AnnotationForge, rcartocolor) all sit in
eval: false or commented-out chunks, so none of them is on an executed path.

Author: GuangchuangYu

.github/workflows/publish-book.yml

@@ -88,6 +88,13 @@ jobs:
           #   europepmc  pmcplot()                         enrichplot-specialized.qmd
           #   ggstar     dotplot.compareClusterResult()    the compareCluster chapters
           #   ggrepel    .manhattanplot_internal()         enrichplot-summary.qmd
+          #   BiasedUrn  ora(..., weight = )               enrichment-network.qmd
+          #
+          # BiasedUrn is enrichit's Suggests, not its Imports, so it is absent
+          # unless named. The weighted-ORA chunk in enrichment-network.qmd calls
+          # ora(weight = w), which stops on check_installed('BiasedUrn') rather
+          # than degrading to unweighted ORA. The book runs the weighted path,
+          # so the package has to be on the runner.
           #
           # ggstar and ggrepel probably arrive transitively today (ggtreeExtra,
           # ggplot2). Listing them costs a few seconds and stops that from silently
@@ -128,7 +135,8 @@ jobs:
             "cran::ggupset",
             "cran::ggridges",
             "cran::shadowtext",
-            "cran::ggforce"
+            "cran::ggforce",
+            "cran::BiasedUrn"
           ))'
 
       - name: Install Bioconductor dependencies (release)
@@ -159,7 +167,8 @@ jobs:
           Rscript -e 'required <- c(
             "org.Hs.eg.db", "fgsea", "GO.db", "ChIPseeker", "MeSHDbi", "biomaRt", "pathview",
             "R.utils", "ggarchery", "europepmc", "ggstar", "ggrepel", "ggupset", "ggridges",
-            "ggplotify", "httr2", "dotenv", "msigdbr", "ggplot2", "dplyr", "stringr", "forcats"
+            "ggplotify", "httr2", "dotenv", "msigdbr", "ggplot2", "dplyr", "stringr", "forcats",
+            "BiasedUrn"
           ); missing <- required[!vapply(required, requireNamespace, logical(1), quietly = TRUE)]; if (length(missing)) stop("Missing packages: ", paste(missing, collapse = ", "))'
 
       # aisdk.providers carries the DeepSeek and other non-bundled adapters;

enrichment-network.qmd

@@ -245,6 +245,22 @@ head(as.data.frame(res_wora)[, c("ID", "SetSize", "Count", "pvalue")])
 head(as.data.frame(res_wgsea)[, c("ID", "setSize", "enrichmentScore", "NES", "pvalue")])
 ```
 
+::: {.callout-note}
+## Weighted ORA needs `BiasedUrn` installed
+
+`BiasedUrn` is a *suggested* package of `enrichit`, so a default install does not
+bring it in. Weighted ORA calls `BiasedUrn::pWNCHypergeo()` and **stops** when it
+is missing — it does not fall back to ordinary ORA:
+
+```r
+Error: The package "BiasedUrn" is required for weighted ORA.
+```
+
+Install it with `install.packages("BiasedUrn")` before running this section.
+Weighted GSEA is unaffected; the weights are fused into the ranked statistics
+before the C++ engine, so no extra package is needed.
+:::
+
 ::: {.callout-warning}
 ## `weight` must share the namespace of the genes