Michael by mnneely · Pull Request #309 · LAPKB/Pmetrics

mnneely · 2026-03-25T18:02:31Z

Number of updates and bug fixes

Add BestDose dose optimization feature ported from Pmetrics_rust bestdose branch: - R/PM_bestdose.R: PM_bestdose and PM_bestdose_problem R6 classes for Bayesian dose optimization with support for concentration and AUC targets - src/rust/src/bestdose_executor.rs: Rust backend for BestDose optimization using pmcore's BestDoseProblem with ODE model support - Updated lib.rs with bestdose, bestdose_prepare, bestdose_optimize exports - Updated extendr-wrappers.R with R-side wrapper functions - Updated NAMESPACE with PM_bestdose, PM_bestdose_problem, bestdose exports - Bumped pmcore dependency from 0.21.1 to 0.22.1 (required for bestdose) - Added libloading dependency for dynamic model loading - Added bestdose example data (past, prior, target CSVs) and test script - Fixed executor.rs mutability issue for pmcore 0.22.1 compatibility

…iates

… dose

Reworked bd and bd_post R6 classes: bd_post$optimize() now returns a bd object, and bd$new() supports one-shot usage. Removed dead Rust code (bestdose(), bestdose_ode(), bestdose_analytical()). Fixed bias_weight being NULL and print() breaking with variable dose counts. Updated PMcore dep to use relative time_offset gap semantics.

- Combine clean one-shot/two-stage API with inline future=list(...) support - Add top-5 weighted plot curves, triangle dose indicators, plot() method - Fix .sim_future() dose recycling when past doses are present

…sed most of the logs for bestdose, needs more work

Copilot

Pull request overview

This PR delivers a broad cleanup + feature update across the R package and Rust backend, including removal of legacy/vignette assets, additions for BestDose reporting, new user options (date formatting), improved model comparison highlighting, and updated Rust logging/dependencies.

Changes:

Adds BestDose HTML report template and expands user-configurable options (notably date_format) plus helpers to fetch/download latest R release.
Updates model comparison “best metric” selection/highlighting and refines simulation/model APIs (quiet, digits handling, arg validation).
Updates Rust backend (pmcore bump, logging defaults, minor executor/settings changes) and removes large amounts of archived/legacy material and old vignette resources.

Reviewed changes

Copilot reviewed 57 out of 88 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
vignettes/Styles/style_old.css	Removed legacy vignette stylesheet.
vignettes/Styles/style.css	Removed vignette stylesheet.
vignettes/Images/unibv40.png	Vignette image asset update.
vignettes/Images/schemaTop.png	Vignette image asset update.
vignettes/Images/rust-logo-64x64.png	Vignette image asset update.
vignettes/Images/pta2.png	Vignette image asset update.
vignettes/Images/pta1.png	Vignette image asset update.
vignettes/Images/plotBrowser.png	Vignette image asset update.
vignettes/Images/multi-bv40.png	Vignette image asset update.
vignettes/Images/model_builder/secondary.png	Vignette image asset update.
vignettes/Images/model_builder/primary.png	Vignette image asset update.
vignettes/Images/model_builder/output2.png	Vignette image asset update.
vignettes/Images/model_builder/output.png	Vignette image asset update.
vignettes/Images/model_builder/lag.png	Vignette image asset update.
vignettes/Images/model_builder/front.png	Vignette image asset update.
vignettes/Images/model_builder/eqn.png	Vignette image asset update.
vignettes/Images/model_builder/covariates.png	Vignette image asset update.
vignettes/Images/installPkg.png	Vignette image asset update.
vignettes/Images/export.png	Vignette image asset update.
vignettes/Images/downloadPNG.png	Vignette image asset update.
vignettes/Images/Slide2.png	Vignette image asset update.
vignettes/Images/Slide1.png	Vignette image asset update.
vignettes/Images/Rlogo.png	Vignette image asset update.
vignettes/Images/Pmetrics_logo.png	Vignette image asset update.
vignettes/Data/reserved.csv	Removed vignette data table.
vignettes/Data/mdata.csv	Removed vignette data example.
vignettes/Data/fortran2.csv	Removed vignette reference table.
vignettes/Data/fortran1.csv	Removed vignette reference table.
vignettes/Data/RLcomp_valid.csv	Removed vignette comparison table.
vignettes/Data/RLcomp_other.csv	Removed vignette comparison table.
vignettes/Data/RLcomp_data.csv	Removed vignette comparison table.
vignettes/Data/R6_Legacy_compare.xlsx	Vignette spreadsheet asset update.
vignettes/Data/PM_result_methods.csv	Removed vignette data table.
vignettes/Data/PM_result_fields.csv	Removed vignette data table.
vignettes/Data/PM_result.xlsx	Vignette spreadsheet asset update.
vignettes/Data/ObjectsR6.csv	Removed vignette data table.
vignettes/Data/ObjectsLegacy.csv	Removed vignette data table.
vignettes/Citations.bib	Removed vignette bibliography file.
src/rust/src/settings.rs	Sets pmcore log level to INFO when writing settings.
src/rust/src/lib.rs	Adjusts tracing subscriber default filter to WARN; formatting/whitespace edits.
src/rust/src/executor.rs	Makes fit result mutable before writing outputs.
src/rust/Cargo.toml	Bumps pmcore to 0.22.1; adds libloading dependency; keeps extendr-api wildcard.
man/summary.PM_sim.Rd	Aligns `digits` default with `getPMoptions("digits")`; doc wording updates.
man/latestR.Rd	New docs for `latestR()`.
man/downloadR.Rd	New docs for `downloadR()`.
man/PMcheck.Rd	Updates `PMcheck()` usage default `path = "."`.
man/PM_model.Rd	Documents new `quiet` parameters for `fit`, `sim`, and `compile`.
man/PM_compare.Rd	Clarifies “best metric” rules in return value description.
inst/report/templates/bestdose.Rmd	Adds a BestDose report template (DT/plotly-based).
inst/options/PMoptions.json	Adds `date_format` default option.
data-raw/data-raw.R	Disables report generation in an example fit (`report = "none"`).
R/zzz.R	Adds interactive attach-time version checks (Pmetrics + R) and related messaging.
R/model_transpiler.R	Improves template mapping/validation; errors for unsupported clearance templates; minor Rust header formatting changes.
R/PMutilities.R	Updates `cli_df` highlighting behavior; adds `latestR()`/`downloadR()` utilities; retains `clear_build()`.
R/PMoptions.R	Adds locale-based `date_format` initialization and Shiny UI control for it.
R/PM_sim.R	Adds `...` arg validation, adjusts parameter counting for theta mode, improves covariate sim summary calcs, changes recompilation behavior, and updates summary docs/behavior.
R/PM_post.R	Drops `pop_mean`/`pop_median` before reshaping to posterior predictions.
R/PM_model.R	Adds `quiet` plumbing into `fit/sim/compile`; changes compilation messaging; alters template_path selection.
R/PM_compare.R	Updates convergence detection and rewrites “best” metric counting + highlight metadata.
NAMESPACE	Exports new `latestR` and `downloadR`.
DESCRIPTION	Bumps package version to 3.0.10.
Archived/summaryPMdopt.R	Removed archived legacy code.
Archived/printPMdopt.R	Removed archived legacy code.
Archived/pmetrics-logo.png	Archived asset update/removal.
Archived/plotPMdopt.R	Removed archived legacy code.
Archived/pkgdown.yml	Removed archived CI config.
Archived/modelLibrary.R	Removed archived model library script.
Archived/makePTA2.R	Removed archived legacy PTA script.
Archived/genAlquimiaData.R	Removed archived legacy script.
Archived/examples_old.R	Removed archived legacy examples.
Archived/Pmetrics_old.css	Removed archived legacy CSS.
Archived/Pmetrics.css	Removed archived legacy CSS.
Archived/PMupdate.R	Removed archived legacy updater script.
Archived/PMsave.R	Removed archived legacy save helper.
Archived/PMreport.R	Removed archived legacy reporting implementation.
Archived/PMremote.R	Removed archived legacy remote server code.
Archived/MMopt.R	Removed archived legacy optimal design implementation.
Archived/LegacyPlots.R	Removed archived legacy plotting code.
Archived/LICENSE	Removed archived license copy.
Archived/ERRrepScript.R	Removed archived legacy script.
Archived/.gitignore	Removed archived .gitignore.
.gitignore	Stops ignoring `inst/Examples/`; adds ignore for `errors.xlsx`; ignores `Archived/`.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 57 out of 89 changed files in this pull request and generated 5 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-25T18:18:17Z

    settings.set_output_path(output_path.to_string());
    settings.set_write_logs(true);
+    settings.set_log_level(LogLevel::INFO);
    settings.write()?;


settings.set_log_level(LogLevel::INFO) hard-codes INFO logging, but setup_logs() now filters WARN+ by default. Consider making the log level configurable (e.g., from the R settings list) and/or setting it to WARN to match the subscriber default, otherwise INFO logs may be written but never surfaced.

Copilot · 2026-03-25T18:18:18Z

+pmcore = { version = "=0.22.1", features = ["exa"] }
+libloading = "0.9"



libloading was added as a dependency but does not appear to be used anywhere in src/rust/src (no references found). Consider removing it to reduce compile time and dependency surface, or add the intended usage so this stays justified.

Copilot · 2026-03-25T18:18:18Z

 fn setup_logs() -> anyhow::Result<()> {
    use tracing::Level;
    use tracing_subscriber::filter::LevelFilter;

    // Create a subscriber with our custom layer using the global timer
-    // Filter to show only INFO and above (INFO, WARN, ERROR)
+    // Filter to show only WARN and above (WARN, ERROR) by default
    let subscriber = tracing_subscriber::registry()
        .with(RFormatLayer::new())
-        .with(LevelFilter::from_level(Level::INFO));
+        .with(LevelFilter::from_level(Level::WARN));



The Rust tracing subscriber is configured to filter at WARN+ by default here, but settings.rs sets Settings log level to INFO. This mismatch can make it look like INFO logging is enabled (and may generate log files) while nothing shows up in R output. Consider aligning the defaults (both WARN or both INFO) or wiring the subscriber filter from the same setting.

Copilot

Pull request overview

Copilot reviewed 59 out of 91 changed files in this pull request and generated 3 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-26T04:37:13Z

+    # Validate update settings
+    valid_update_check <- c("manual", "never", "daily", "weekly", "monthly")
+    update_check <- tolower(as.character(synced_opts$update_check))
+    if (!(update_check %in% valid_update_check)) {
+      update_check <- "weekly"
+    }
+    synced_opts$update_check <- update_check


The UI validation for update_check does not include the value "always", but pm_update_interval_days() supports "always" as a valid mode. This makes "always" effectively impossible to persist via the options UI and leaves dead/inconsistent behavior. Either add "always" to the validated choices (and optionally expose it in the UI) or remove it from pm_update_interval_days() to keep the option surface consistent.

Copilot · 2026-03-26T04:37:14Z

+      metric_best <- purrr::map(metric_cols, function(col) {
+        x_num <- suppressWarnings(as.numeric(df[[col]]))
+        valid <- which(!is.na(x_num))
+        if (length(valid) == 0) return(integer(0))
+        target <- ifelse(grepl("Sl|R2", col), 1, 0)
+        valid[which(abs(x_num[valid] - target) == min(abs(x_num[valid] - target)))]
+      })


The "best"-value highlighting logic treats all non-slope/non-R2 metrics as having an ideal target of 0 (i.e., chooses values closest to 0). This is incorrect for metrics like -2*LL/AIC/BIC where the best value should be the minimum (most negative/lowest), not the closest to 0. Consider special-casing LL/AIC/BIC to use min(x) (and slope/R2 to use closest to 1, bias/imprecision/intercept to use closest to 0) so the CLI highlighting matches the documented metric semantics.

Copilot · 2026-03-26T04:37:14Z

+  target <- ifelse(grepl("Sl|R2", col), 1, 0)
+  valid[which(abs(x[valid] - target) == min(abs(x[valid] - target)))]


The computation of per-metric "best" rows uses a closest-to-target (0 or 1) heuristic for all metric columns. This conflicts with the function documentation for -2*LL/AIC/BIC, where the best model is the one with the lowest value, not the value closest to 0. This can mis-rank models (especially if any of these metrics can be negative). Suggest explicitly selecting minima for ll/aic/bic, closest-to-0 for bias/imprecision/intercepts, and closest-to-1 for slopes/R2 so both the best-counts and highlighting match the documented behavior.

Suggested change

target <- ifelse(grepl("Sl|R2", col), 1, 0)

valid[which(abs(x[valid] - target) == min(abs(x[valid] - target)))]

# Determine "best" rows per metric:

# - For -2*LL/AIC/BIC, best is the minimum value.

# - For slopes and R2, best is closest to 1.

# - For bias, imprecision, intercepts, best is closest to 0.

if (grepl("LL$|AIC$|BIC$", col, ignore.case = TRUE)) {

# Information criteria / likelihood: smaller is better

best_val <- min(x[valid])

valid[which(x[valid] == best_val)]

} else if (grepl("Sl|R2", col, ignore.case = TRUE)) {

# Slopes and R2: closer to 1 is better

diffs <- abs(x[valid] - 1)

best_diff <- min(diffs)

valid[which(diffs == best_diff)]

} else {

# Bias, imprecision, intercepts: closer to 0 is better

diffs <- abs(x[valid])

best_diff <- min(diffs)

valid[which(diffs == best_diff)]

}

mnneely · 2026-03-26T14:32:07Z

@copilot open a new pull request to apply changes based on the comments in this thread

Copilot · 2026-03-26T14:32:48Z

@mnneely I've opened a new pull request, #311, to work on those changes. Once the pull request is ready, I'll request review from you.

Siel and others added 30 commits February 12, 2026 09:03

fix: use devtools::load_all() in bestdose example script

04f741b

fix: limits from model not applied properly

19b924c

Merge remote-tracking branch 'origin/main' into Michael

e0a7bd9

fix: malformed PM_post

9cf18bb

fix: plotting post predictions with data; remade data objects

3602e9b

fix: inappropriate truncating of long color names in plot.PM_data

76da1a4

wip: update summary.PM_sim

1f49bbe

fix: max_cycles are passed correctly, also error model

44b0d6a

fix: missing path in PMcheck or misspelled arguments in PM_sim

377ab05

feat: new bestdose API

9fc29dd

chore: remove old functions

f909cc4

fix: using PMcore on gh

7c8e433

chore: remove optimize-raw

817cfeb

Merge remote-tracking branch 'origin/bestdose' into BD-Michael

e6cf5df

chore: simplify function names

c884649

Merge remote-tracking branch 'origin/Michael' into BD-Michael

898200f

chore: update documentation after merge

3f3910a

wip: creating simulated past

8a44db7

wip: update example objects

9ff94b5

wip: documentation

0b62a5f

wip: compile with silence and only recompile if simulating with covar…

dc79248

…iates

feat: add plots

fca65d7

fix: remove unneeded example

c9b06fb

wip: register S3 plot.bd method

bd07b05

wip: update plot, add inline targets, sort standard data by id, time,…

47fd9b5

… dose

Merge BD-Michael: integrate inline future API and improved plotting

3ac3e99

- Combine clean one-shot/two-stage API with inline future=list(...) support - Add top-5 weighted plot curves, triangle dose indicators, plot() method - Fix .sim_future() dose recycling when past doses are present

fix: fixed some errors in the model definition of the example, suppre…

c2e20c9

…sed most of the logs for bestdose, needs more work

fix: the bd_new example now has no time_offset

f2fc5df

mnneely added 9 commits March 16, 2026 15:51

fix: update best logic in PM_compare

e89f508

fix: remove errors file

f9be770

wip: update documentation for PM_compare

a288cc6

wip: removing bestdose artifacts from this version

132a268

fix: notify user cl models not yet supported

f31d8f5

fix: handle missing analytic models

7fb037d

fix: update across function

5a5a60a

wip: cleaning old files

93df243

Stop tracking Archived; add to .gitignore

2293826

Copilot AI review requested due to automatic review settings March 25, 2026 18:02

Copilot started reviewing on behalf of mnneely March 25, 2026 18:02 View session

Merge remote-tracking branch 'origin/main' into Michael

ca890b2

Copilot AI reviewed Mar 25, 2026

View reviewed changes

Comment thread R/PM_model.R

Comment thread R/zzz.R Outdated

Comment thread R/zzz.R Outdated

Comment thread R/PM_sim.R Outdated

Comment thread R/PM_sim.R Outdated

mnneely and others added 2 commits March 25, 2026 11:09

wip: update objects

fa21ba3

Update R/PM_sim.R

f6f1a8b

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

Copilot AI review requested due to automatic review settings March 25, 2026 18:11

Copilot started reviewing on behalf of mnneely March 25, 2026 18:12 View session

mnneely added 2 commits March 25, 2026 11:16

fix: typo in help

906b4b2

Merge branch 'Michael' of https://github.com/LAPKB/Pmetrics into Michael

b888476

Copilot AI reviewed Mar 25, 2026

View reviewed changes

mnneely added 2 commits March 25, 2026 21:24

fix: check for updates periodically

133f853

fix: add check_updates, latestR, downloadR to _pkgdown.yml

ab132e5

Copilot AI review requested due to automatic review settings March 26, 2026 04:33

Copilot started reviewing on behalf of mnneely March 26, 2026 04:33 View session

Copilot AI reviewed Mar 26, 2026

View reviewed changes

Copilot AI mentioned this pull request Mar 26, 2026

Fix best-metric highlighting for LL/AIC/BIC and add "always" to update_check options #311

Draft

Siel approved these changes Mar 26, 2026

View reviewed changes

mnneely merged commit 369d7c1 into main Mar 26, 2026
15 checks passed

		pmcore = { version = "=0.22.1", features = ["exa"] }
		libloading = "0.9"

		target <- ifelse(grepl("Sl\|R2", col), 1, 0)
		valid[which(abs(x[valid] - target) == min(abs(x[valid] - target)))]

-  target <- ifelse(grepl("Sl|R2", col), 1, 0)
-  valid[which(abs(x[valid] - target) == min(abs(x[valid] - target)))]
+  # Determine "best" rows per metric:
+  # - For -2*LL/AIC/BIC, best is the minimum value.
+  # - For slopes and R2, best is closest to 1.
+  # - For bias, imprecision, intercepts, best is closest to 0.
+  if (grepl("LL$|AIC$|BIC$", col, ignore.case = TRUE)) {
+    # Information criteria / likelihood: smaller is better
+    best_val <- min(x[valid])
+    valid[which(x[valid] == best_val)]
+  } else if (grepl("Sl|R2", col, ignore.case = TRUE)) {
+    # Slopes and R2: closer to 1 is better
+    diffs <- abs(x[valid] - 1)
+    best_diff <- min(diffs)
+    valid[which(diffs == best_diff)]
+  } else {
+    # Bias, imprecision, intercepts: closer to 0 is better
+    diffs <- abs(x[valid])
+    best_diff <- min(diffs)
+    valid[which(diffs == best_diff)]
+  }

Conversation

mnneely commented Mar 25, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Mar 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 26, 2026

Choose a reason for hiding this comment

Uh oh!

mnneely commented Mar 26, 2026

Uh oh!

Copilot AI commented Mar 26, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants