Model description and upcoming additions by Kwan-Jenny · Pull Request #88 · UCD-SERG/serodynamics

Kwan-Jenny · 2025-04-15T00:44:08Z

This pull request adds the finalized Beamer presentation materials for the April 14th meeting, including:

Beamer_Antibody_Kinetics.qmd: Full 32-slide presentation in Quarto Beamer format
Beamer_Antibody_Kinetics.pdf: Compiled version for presentation
Slides incorporate feedback from Dr. Morrison on the antibody kinetics model in Chapter 2
Covers model equations, parameter interpretations, hierarchical prior structures, and proposed correlation modeling across biomarkers

Please let me know if you want me to change anything in here.

codecov · 2025-04-15T00:49:39Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ All tests successful. No failed tests found.

github-actions · 2025-04-15T03:34:38Z

📖 https://ucd-serg.github.io/serodynamics/preview/pr88
Preview documentation for this PR (at commit b18c735)

d-morrison

Looks good so far! See comments

d-morrison · 2025-04-15T02:18:08Z

Please don't track the pdf version

# add references bib file # Remove tracked PDF and .tex files and add to .gitignore

…rmat

# spelling correction # update WORDLIST

Merge branch 'main' of https://github.com/UCD-SERG/serodynamics into beamer # Conflicts: # .gitignore

Kwan-Jenny · 2025-05-30T01:05:04Z

Currently, there are three .qmd files in the articles folder, but Antibody_Kinetics.qmd is the most up-to-date version. Therefore, only Antibody_Kinetics.qmd needs to be reviewed. However, I am unsure whether I should delete the other two .qmd files. @d-morrison, should I delete the other two and keep only this one?

sschildhauer

This looks great Kwan! Just a few small clarifications. After reviewing you can re request from me and I will try to review in a more timely manner.

sschildhauer · 2025-05-30T00:25:11Z

@@ -0,0 +1,250 @@
+---
+title: "Hierarchical Bayesian Model"
+author: "Our Study Group"


Suggested change

author: "Our Study Group"

author: "Kwan Ho Lee, UC Davis SeroEpidemiology Research Group"

sschildhauer · 2025-05-30T00:28:21Z

+
+## Big Picture: What Are We Modeling?
+
+We are modeling **how antibody levels change over time** in response to infection, using data from multiple individuals and multiple **biomarkers** (10 antigen-isotype combinations, so ( j = 1, 2, ..., 10 )).


Suggested change

We are modeling **how antibody levels change over time** in response to infection, using data from multiple individuals and multiple **biomarkers** (10 antigen-isotype combinations, so ( j = 1, 2, ..., 10 )).

We are modeling **how antibody levels change over time** in response to infection for different antigen-isotype (biomarker) combinations (ex. anti-LPS IgG), using longitudinal serologic data from multiple individuals (10 antigen-isotype combinations, so ( j = 1, 2, ..., 10 )).

sschildhauer · 2025-05-30T00:29:01Z

+
+## Big Picture: What Are We Modeling?
+
+We are modeling **how antibody levels change over time** in response to infection, using data from multiple individuals and multiple **biomarkers** (10 antigen-isotype combinations, so ( j = 1, 2, ..., 10 )).


Why is this 10 antigen-isotype combinations? Couldn't it technically be any number of combinations?

sschildhauer · 2025-05-30T00:38:59Z

+We want to:
+
+-   Understand the average pattern for each biomarker
+-   Allow each person’s response to vary


Suggested change

- Allow each person’s response to vary

- Allow each person’s response to vary as a random effect

sschildhauer · 2025-05-30T00:41:42Z

+\end{bmatrix}
+$$
+
+These describe the antibody curve for person ( $i$ ) and biomarker ( $j$ ): the starting level, how fast it rises, peaks, and decays.


Suggested change

These describe the antibody curve for person ( $i$ ) and biomarker ( $j$ ): the starting level, how fast it rises, peaks, and decays.

These describe the antibody kinetic curve for each person ( $i$ ) and biomarker ( $j$ ): the baseline antibody level, rate of increase, peaks antibody level, and decay (waning) rate.

sschildhauer · 2025-06-01T17:05:52Z

+
+Higher $\nu_j$ $\rightarrow$ more informative prior (stronger prior).
+
+Lower $\nu_j$ $\rightarrow$ more weakly informative (broader prior or weaker prior).


Suggested change

Lower $\nu_j$ $\rightarrow$ more weakly informative (broader prior or weaker prior).

Lower $\nu_j$ $\rightarrow$ less informative (broader prior or weaker prior).

sschildhauer · 2025-06-01T17:08:38Z

+
+The model is built hierarchically across five conceptual levels:
+
+1.  **Observed data:** noisy log antibody concentrations from serum samples


Suggested change

1. **Observed data:** noisy log antibody concentrations from serum samples

1. **Observed data:** log antibody concentrations from serum samples

sschildhauer · 2025-06-01T17:09:54Z

+The model is built hierarchically across five conceptual levels:
+
+1.  **Observed data:** noisy log antibody concentrations from serum samples
+2.  **Latent individual parameters:** hidden antibody dynamics $\theta_{ij}$ for each subject-biomarker pair


What does "hidden" mean here? Does it mean unobserved?

sschildhauer · 2025-06-01T17:11:17Z

+
+2.  **Middle Level**:
+
+    -   For each person $i$, their parameters:


Suggested change

- For each person $i$, their parameters:

- Each person $i$ has their parameters:

sschildhauer · 2025-06-01T17:12:14Z

+
+3.  **Bottom Level**:
+
+    -   Their actual observed antibody levels are noisy measurements of predictions from $\theta_{ij}$:


Suggested change

- Their actual observed antibody levels are noisy measurements of predictions from $\theta_{ij}$:

- Observed antibody levels are noisy measurements of predictions from $\theta_{ij}$:

Kwan-Jenny · 2026-05-21T18:06:20Z

@copilot resolve the merge conflicts in this pull request

Co-authored-by: Kwan-Jenny <68584166+Kwan-Jenny@users.noreply.github.com>

Copilot · 2026-05-21T18:13:59Z

@copilot resolve the merge conflicts in this pull request

Resolved — I merged origin/main into this PR branch and fixed the conflicts in .gitignore, DESCRIPTION, and NEWS.md in commit 5a1823b.

Kwan-Jenny · 2026-05-21T18:39:46Z

@claude review

claude · 2026-05-21T18:43:08Z

Claude finished @Kwan-Jenny's task in 2m 23s —— View job

Claude PR Review

Read repo structure
Examine changed files (git diff origin/main...HEAD)
Review vignettes/articles/Antibody_Kinetics.qmd
Review vignettes/articles/Beamer_Antibody_Kinetics.qmd
Review vignettes/articles/Hierarchical_Bayesian_Model.qmd
Check tracked PDF / .tex files against prior review feedback
Check .gitignore, NEWS.md, inst/WORDLIST, references.bib
Post review feedback

Summary

The presentation/article content has come a long way and the notation cleanup ($\mu_y$, $\mu_b$, $\gamma$, $\rho$, citation system) addresses most of @d-morrison's earlier requests. But several blocking items from the prior reviews are still unresolved, plus a couple of new bugs introduced by recent edits. Notes below in priority order.

Blocking

1. PDFs and .tex files are still being tracked. @d-morrison asked twice not to track these. They are still in the tree:

vignettes/Beamer_Antibody_Kinetics.pdf (77 KB)
vignettes/articles/Antibody_Kinetics.pdf (79 KB)
vignettes/articles/Antibody_Kinetics.tex (33 KB, 985 lines)

The existing .gitignore rules vignettes/*.pdf and vignettes/*.tex only match the top-level vignettes/ directory; nothing covers vignettes/articles/. To fix: git rm the three files and extend the ignore rules, e.g. add:

vignettes/articles/*.pdf
vignettes/articles/*.tex

2. .gitignore contains a broken line introduced in the merge. Line 14 currently reads docsvignettes/*.pdf — that is not a valid path (it matches nothing). It also looks like docs/ was removed but docs (line 18) is still there, so the intent is unclear. Please fix this typo as part of the same edit.

3. PR scope is ambiguous — three .qmd files, only one is current. Per Jenny's own comment, only Antibody_Kinetics.qmd is up to date, yet the PR is adding Beamer_Antibody_Kinetics.qmd and Hierarchical_Bayesian_Model.qmd as new files. If those two are superseded, please delete them from this PR rather than landing three overlapping documents. If they are intentionally kept as separate artifacts (slides vs. article vs. study-group walkthrough), please add a short note in NEWS.md/the article frontmatter clarifying which is which — pkgdown will render all three as separate articles otherwise.

4. Bibliography is incomplete relative to @d-morrison's request. The review asked to summarize Teunis 2016 and 2023 models before introducing the revised model. Neither file does this, and references.bib only contains teunis2016 — there is no teunis2023 entry at all. Either add a Teunis 2023 entry and a summary slide/section, or note explicitly that this will be added in a follow-up.

5. references.bib has no trailing newline — minor, but several tools (and usethis::use_tidy_description) will flag this. Add a final \n.

Math / content bugs

6. Hierarchical_Bayesian_Model.qmd:79 — wrong dimensions for $\Sigma_j$.

$\Sigma_j$ : $j×j$ covariance matrix describing how the parameters co-vary

This is a mathematical error: $\Sigma_j$ is a covariance matrix over the 7 subject-level parameters, so it is $7\times 7$ (not $j\times j$, which would make it $1\times 1$ when $j=1$). Compare with Antibody_Kinetics.qmd:225 and Beamer_Antibody_Kinetics.qmd:308, which both correctly state $\Sigma_j \in \mathbb{R}^{7\times 7}$.

7. Antibody_Kinetics.qmd:133 and :180 — escaped > will not render.

-   For $t \> t_1$: $$

\> is not valid LaTeX/Quarto math; it renders literally. Should be $t > t_1$ . The earlier ODE block at line 116 uses the correct form.

8. Antibody_Kinetics.qmd:247 and :416 — stray backslash-escaped underscores break LaTeX.

-   These 5 parameters are **log-transformed** into the latent parameters $\theta\_{ij}$ used for modeling.

1.  $\theta_{ij}$ are the **subject-level latent parameters** (like $y_0, b_0, \mu\_b, \mu\_y, \gamma, \alpha, \rho$).

Inside $...$ the underscores should be plain _, not \_. As written, this renders the literal text \_ instead of subscripting.

9. Beamer_Antibody_Kinetics.qmd:412-427 — old notation leaks through. The "Understanding $\text{vec}(\Theta_i)$" slide still lists the vector entries as $y_0, b_0, \mu_0, \mu_1, c, \alpha, r$ even though every other slide in the deck now uses $\mu_b, \mu_y, \gamma, \alpha, \rho$. Please update so the notation is consistent throughout.

10. Antibody_Kinetics.qmd YAML defines beamer: AND pdf:. That produces two PDFs on render (Beamer slides + an article PDF). Given that the article version is already separately delivered via html: default, the second pdf: block is probably unintended — and it is what makes Antibody_Kinetics.pdf and .tex keep showing up in the working tree. Either drop the pdf: block, or — if you actually want both — at minimum get the resulting files out of git per point (1).

Outstanding from prior reviewers (still applies)

These were requested but not yet addressed:

@d-morrison, title suggestion for the article: "Extending the Hierarchical Model for Antibody Kinetics" — the Beamer deck adopted it, but Antibody_Kinetics.qmd:2 still says "Hierarchical Bayesian Modeling of Antibody Kinetics: Extensions and Refinements". Pick one consistently.
@sschildhauer on hardcoded 10: still present in Antibody_Kinetics.qmd:39 ("(j = 1, 2, $\dots$, 10)") — generalize to $n$ or state the constraint.
@sschildhauer on Antibody_Kinetics.qmd:63 ($\theta$ not present in the equation above): still applies — either reference $\theta_{ij}$ explicitly in eq-1 or remove the bullet.
@sschildhauer on Antibody_Kinetics.qmd:111: still no sentence explaining how the ODE system relates to the closed-form solutions that follow.
@sschildhauer on Antibody_Kinetics.qmd:379 ("hidden"): the new wording at line 379 still uses "hidden antibody dynamics" — replace with "unobserved/latent" for clarity.

I'd suggest going through @sschildhauer's review systematically and either applying the suggestion-block edits or replying inline with the reasoning where you disagree.

Smaller nits

Beamer_Antibody_Kinetics.qmd:22 — Dr.Aiemjoy is missing a space (Dr. Aiemjoy).
Hierarchical_Bayesian_Model.qmd:3 — author is "Our Study Group". If this slide deck is staying in the PR, please set a real author (or remove the file per point 3).
inst/WORDLIST contains a number of duplicate entries (biomarker, biomarkers, isotype, isotypes, mcmc, nmc, vec, hyperprior, Teunis, hyp). spelling::update_wordlist() will de-duplicate and sort it for you.
inst/WORDLIST also adds plain math/LaTeX command names (bmatrix, boldsymbol, cdot, cdots, dt, dy, frac, ge, le, leq, mathbb, mathcal, mathrm, otimes, rightarrow, et, th, cd). These suggest the spellchecker is being run on the rendered output and picking up TeX tokens; it might be cleaner to exclude the .qmd/rendered HTML from spellcheck rather than allowlisting every macro name.
NEWS.md:50 — entry currently reads "Added Beamer presentation files (Beamer_Antibody_Kinetics.qmd, .pdf) (Model description and upcoming additions #88)". If you stop tracking the .pdf per point 1 (and if you keep Antibody_Kinetics.qmd as the canonical doc per point 3), update this line to match what the PR actually ships.

Things that look good

Notation harmonization to $\mu_y, \mu_b, \gamma, \rho$ across files.
[@teunis2016] Quarto citation usage in both Antibody_Kinetics.qmd and Beamer_Antibody_Kinetics.qmd.
Adding html: default so pkgdown picks these up as articles (per @d-morrison's earlier comment).
Cross-references between equations and model.jags line ranges in Antibody_Kinetics.qmd are very helpful for a reader trying to map math → code.

I'm posting this as review feedback only — I have not made any changes to the branch. Happy to apply a subset of these if you'd like; just reply with which ones (e.g. "untrack the PDFs and .tex, fix the .gitignore typo, and fix items 6–9").
• Branch: beamer

04.14.25 meeting material

a3e67fb

Kwan-Jenny requested a review from d-morrison April 15, 2025 00:44

# update version

df79edd

Kwan-Jenny and others added 5 commits April 14, 2025 17:51

update NEWS.md file

6d56eff

update WORDLIST

f767fcb

update WORDLIST one more time

ce49627

add Beamer in WORDLIST

485fd23

Merge branch 'main' into beamer

f67dbc8

d-morrison changed the title ~~04.14.25 Meeting Material: Beamer Presentation (Chapter 2 Model Revisions)~~ Model description and upcoming additions Apr 15, 2025

d-morrison requested changes Apr 15, 2025

View reviewed changes

Kwan-Jenny and others added 8 commits April 29, 2025 09:45

# Few updates based on Dr.Morrison's suggestion

cd362b3

# update overview slide

22917f9

# update model structure

698acbe

# revised slides until Parameter summary

13acfcf

# update qmd file

78277da

# add references bib file # Remove tracked PDF and .tex files and add to .gitignore

Merge branch 'main' into beamer

c32de0b

# update version

b58ee62

Resolve merge conflict in DESCRIPTION

b276508

Kwan-Jenny requested a review from d-morrison April 29, 2025 23:47

Kwan-Jenny added 9 commits April 29, 2025 16:53

# Update gitignore to untrack docs folder

8da051d

Stop tracking docs/ folder; added to .gitignore

76c7930

# Add le for spellcheck

82e9877

# code cleaning

ff4d4ac

# cleaning again

396534e

# turns off that caching behavior

5aadf19

# add 1 caption to table

ef2fe38

# cleaning one more time

f315d30

# one more caption and 2 equations numbering

5ebc11a

Kwan-Jenny added 3 commits May 1, 2025 12:29

# add explanatoin about Hierarchical bayesian model in .qmd beamer fo…

f20bc2c

…rmat

# update WORDLIST

1681bcb

# typo correct

97f986d

Kwan-Jenny removed the request for review from d-morrison May 7, 2025 01:07

Kwan-Jenny and others added 3 commits May 15, 2025 15:18

# Update version based on 05.08.25 meeting

436775c

Merge branch 'main' into beamer

a3dfa1d

# update version

f0d26af

# spelling correction # update WORDLIST

Kwan-Jenny requested a review from sschildhauer May 16, 2025 01:04

Kwan-Jenny added 7 commits May 22, 2025 09:55

# was edited by last meeting but will be fixed soon

2e74a54

# pull main branch based on PR #100

f02de1c

Merge branch 'main' of https://github.com/UCD-SERG/serodynamics into beamer # Conflicts: # .gitignore

# move .qmd files and .bib file under articles folder

79e8e8f

# revised qmd file based on Dr.Morrison's suggestion

f83de69

# update version

e5eb988

# update model.jags code into qmd file

d162e57

# clarify some notations and add table

b18c735

Kwan-Jenny requested review from sschildhauer and removed request for sschildhauer May 22, 2025 23:27

sschildhauer requested changes Jun 1, 2025

View reviewed changes

Kwan-Jenny added 2 commits August 13, 2025 16:24

"WIP: save current local changes"

e5668a3

# revised yaml

05ab5a8

Copilot started work on behalf of Kwan-Jenny May 21, 2026 18:06 View session

Merge origin/main into beamer and resolve conflicts

5a1823b

Co-authored-by: Kwan-Jenny <68584166+Kwan-Jenny@users.noreply.github.com>

Copilot AI review requested due to automatic review settings May 21, 2026 18:13

Kwan-Jenny review requested due to automatic review settings May 21, 2026 18:13

Copilot finished work on behalf of Kwan-Jenny May 21, 2026 18:14

	author: "Our Study Group"
	author: "Kwan Ho Lee, UC Davis SeroEpidemiology Research Group"


		## Big Picture: What Are We Modeling?

		We are modeling how antibody levels change over time in response to infection, using data from multiple individuals and multiple biomarkers (10 antigen-isotype combinations, so ( j = 1, 2, ..., 10 )).

	- Allow each person’s response to vary
	- Allow each person’s response to vary as a random effect

	These describe the antibody curve for person ( $i$ ) and biomarker ( $j$ ): the starting level, how fast it rises, peaks, and decays.
	These describe the antibody kinetic curve for each person ( $i$ ) and biomarker ( $j$ ): the baseline antibody level, rate of increase, peaks antibody level, and decay (waning) rate.


		Higher $\nu_j$ $\rightarrow$ more informative prior (stronger prior).

		Lower $\nu_j$ $\rightarrow$ more weakly informative (broader prior or weaker prior).

	Lower $\nu_j$ $\rightarrow$ more weakly informative (broader prior or weaker prior).
	Lower $\nu_j$ $\rightarrow$ less informative (broader prior or weaker prior).


		The model is built hierarchically across five conceptual levels:

		1. Observed data: noisy log antibody concentrations from serum samples

	1. Observed data: noisy log antibody concentrations from serum samples
	1. Observed data: log antibody concentrations from serum samples


		2. Middle Level:

		- For each person $i$, their parameters:

	- For each person $i$, their parameters:
	- Each person $i$ has their parameters:


		3. Bottom Level:

		- Their actual observed antibody levels are noisy measurements of predictions from $\theta_{ij}$:

	- Their actual observed antibody levels are noisy measurements of predictions from $\theta_{ij}$:
	- Observed antibody levels are noisy measurements of predictions from $\theta_{ij}$:

Conversation

Kwan-Jenny commented Apr 15, 2025

Uh oh!

codecov Bot commented Apr 15, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

github-actions Bot commented Apr 15, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

d-morrison left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Kwan-Jenny commented May 30, 2025

Uh oh!

sschildhauer left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Kwan-Jenny commented May 21, 2026

Uh oh!

Copilot AI commented May 21, 2026

Uh oh!

Kwan-Jenny commented May 21, 2026

Uh oh!

claude Bot commented May 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Claude PR Review

Summary

Blocking

Math / content bugs

Outstanding from prior reviewers (still applies)

Smaller nits

Things that look good

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

codecov Bot commented Apr 15, 2025 •

edited

Loading

github-actions Bot commented Apr 15, 2025 •

edited

Loading

claude Bot commented May 21, 2026 •

edited

Loading