Skip to content

Grice Cluster Review — CSH-016, CSH-017, CSH-001

Branch: development Date: 2026-06-13 Scope: Evaluate whether the Grice Laboratory (CSH-016), Karver Laboratory (CSH-017), and Chemistry Core Instrumentation & Machine Shop (CSH-001) pages adequately represent the institutional knowledge already present in the repository. Constraint: This is an evaluation only. No pages, YAML, schemas, or exports were changed. The goal is evidence-backed refactoring of existing knowledge — not discovering new facts. Related: Development Branch Charter · Main Evidence & Verification Audit · Source Packet Workflow

What was inspected

Page File
Grice Laboratory resources/CSH/CSH-016.md
Karver Laboratory resources/CSH/CSH-017.md
Chemistry Core resources/CSH/CSH-001.md

Supporting evidence reviewed (already in repo):

  • data/people.yaml — PER-100 (Grice), PER-074 (Karver), associations with evidence_text
  • data/person_resource_links.yaml — derived person↔resource edges
  • data/grants.yaml, data/grant_resource_links.yaml — GRA-025/GRL-001, GRA-033/GRL-012
  • data/publications.yaml, data/publication_resource_links.yaml — PUB-009, PUB-013
  • data/resource_access.yaml, data/resource_verification.yaml, data/resource_equipment.yaml
  • Neighbor pages: CSH-018.md (Southern), CSH-003.md (McGowan), CSH-002.md, docs/units/csh.md

Existing strengths

  1. All three pages are evidence-anchored. Each has structured Access, Verification, and Equipment sections sourced from YAML, with rationale and source URLs.
  2. People are well-modeled. people.yaml carries rich evidence_text per association (e.g. Karver "McGowan South 321D office and 325 lab; medicinal and chemical biology of inflammatory caspases"; Grice "Office McGowan South 321C; NMR, IR, UV-Vis, electrochemistry, computational chemistry").
  3. Core reciprocity already works for the main edges:
  4. CSH-016 ↔ CSH-017 (reciprocal)
  5. CSH-016 ↔ CSH-001, CSH-017 ↔ CSH-001 (reciprocal; CSH-001 lists both back)
  6. CSH-016 ↔ CSH-003, CSH-017 ↔ CSH-003 (reciprocal)
  7. CSH-017 ↔ CSH-018 (reciprocal)
  8. Karver's dual role is captured — Faculty lead of CSH-017 and Director (chair) of CSH-001/002/003, with catalog evidence in people.yaml.
  9. Conservative, honest statuses — both labs are likely (~) with mid-80s confidence; access is correctly yellow (affiliation-based, not public booking).
  10. Equipment ties to the shared catalog via category anchors and search hints.

These are hyperlinks that could exist because the target already exists in the repo. None require new facts.

A. Prose mentions that are not hyperlinked

Page Plain-text mention Target exists Currently linked?
CSH-016 Notes "collaborates with Karver Lab" CSH-017 In Linked IDs, not inline
CSH-016 Documented Relationships "shared McGowan core instrumentation (CSH-001)" CSH-001 In Linked IDs, not inline
CSH-017 Documented Relationships "Southern Lab (CSH-018)" CSH-018 In Linked IDs, not inline
CSH-017 Documented Relationships "directs CSH-001/002/003" CSH-001/002/003 CSH-002 not in Linked IDs
CSH-001 Documented Relationships "chemistry, physics, environmental science…" several narrative only
  • CSH-017 → CSH-002 missing. The CSH-017 prose explicitly says the chair "directs CSH-001/002/003," and people.yaml carries director evidence for CSH-002 (PER-074 → CSH-002, catalog source). Yet CSH-002 is absent from CSH-017's ## Linked Relationship IDs. This is an internally-supported link that is simply not surfaced.

C. Publication references with no destination

  • CSH-016 cites "PUB-009 electrocatalysis publication" in its Verification evidence, but the reference is a bare ID — no title, no DOI link. The DOI already exists in data/publications.yaml (10.1021/acs.inorgchem.0c02579, Methane Generation from CO₂ with a Molecular Rhenium Catalyst, Grice coauthor). A reader cannot navigate to it.
Edge CSH-016 CSH-017 CSH-018 Reciprocal?
016 ↔ 017 →017 →016 — Yes
017 ↔ 018 — →018 →017 Yes
016 ↔ 018 — (016, 018) — No

Finding: CSH-016 (Grice) and CSH-018 (Southern) do not link to each other, although both sit in the caspase collaboration cluster that CSH-017 bridges. Caution: the direct Grice↔Southern edge is not strongly evidenced in the repo today — Grice's caspase evidence is with the Karver group (J. Mol. Struct. 2024), and PUB-013 is Karver+Southern. Adding a 016↔018 link would risk asserting a relationship the evidence does not directly support. See Risks. This is listed as a candidate to evaluate, not a confirmed missing link.

Missing evidence visibility

Evidence that exists elsewhere in the repository but is not surfaced where a reader (or reviewer) would look.

  1. PUB-013 is invisible on CSH-017. publication_resource_links.yaml records PUB-013 (Design and Application of a Fluorogenic Assay for Monitoring Inflammatory Caspase Activity, DOI 10.1016/j.ab.2017.11.035) as produced_by CSH-017 with include_in_graph: true, confidence high. This is the single strongest piece of evidence directly matching the lab's stated capability ("fluorogenic peptide-substrate assays; FRET-based caspase activity monitoring"), yet it appears nowhere on the CSH-017 page.
  2. PUB-009 is referenced but not resolvable on CSH-016 (see Missing links C). Note its resource link is include_in_graph: false (method-alignment inference), so it should be presented as supporting, not definitive.
  3. Grants are entirely hidden (by design, but worth stating):
  4. CSH-017 ← GRA-025 Proteomic Analysis of a Covalent Caspase-1 Inhibitor (GRL-001) and an internal URC competitive research grant.
  5. CSH-016 ← GRA-033 URC Paid Leave (Fall 2021) (GRL-012). These derive from internal CSV (data/Internal Grants Summary…). They should stay off public pages per the internal-data permission rules — flagged here only so reviewers know the evidence exists.
  6. Person evidence_text is not shown. The page renders only name/role/email/profile. The richer location/method evidence in people.yaml is available to graph export but not to a human reading the page or reviewing a diff.

Relationship claims that exist only in prose

Page Prose claim Supporting evidence already in repo Could be strengthened by
CSH-016 Notes "collaborates with Karver Lab on caspase inhibitor work" people.yaml PER-100→CSH-017: "Coauthored caspase-1 inhibitor structural study with Karver group (J. Mol. Struct. 2024)" Citing that coauthorship / making it a documented relationship
CSH-017 Documented Relationships "joint FRET-substrate projects with Southern Lab (CSH-018)" PUB-013 (Karver + Southern), include_in_graph: true Citing PUB-013 / DOI
CSH-017 Documented Relationships "Department chair also directs CSH-001/002/003 administration" people.yaml PER-074 director edges, catalog source Linking CSH-002 and citing catalog
CSH-016 Documented Relationships "shared McGowan core instrumentation (CSH-001)" people.yaml PER-100→CSH-001 affiliated evidence Inline link + existing source

All of the above are already supported by repository evidence; the gap is surfacing/citing, not discovering.

Potential low-risk improvements

These improve human navigation and provenance visibility without introducing new factual claims:

  1. Inline-link resource IDs in prose. Where CSH-001, CSH-017, CSH-018 appear in Description / Documented Relationships / Notes, link them to their pages. (Targets already exist; purely navigational.)
  2. Add CSH-002 to CSH-017's Linked Relationship IDs. Supported by existing director evidence; corrects a stated-but-unlinked claim.
  3. Make publication references resolvable. Convert "PUB-009" on CSH-016 and add PUB-013 on CSH-017 to DOI links using DOIs already in publications.yaml. Present PUB-009 as supporting (inferred) and PUB-013 as direct (produced_by).
  4. Cite the Grice–Karver coauthorship for the CSH-016 caspase-collaboration note, using the citation already in people.yaml.
  5. (Optional) Surface key person evidence_text at least in review artifacts so diff reviewers can see provenance without opening raw YAML.
  6. (Cosmetic) Retire the empty ## Source URL boilerplate ("see Contacts above") — it adds no information; lower priority.

Potential risks

  1. Asserting Grice↔Southern (CSH-016↔CSH-018). Tempting for symmetry, but not directly evidenced. Adding it would manufacture a relationship. Do not add without its own source.
  2. Surfacing grants on public pages. Grant evidence derives from internal CSV; repository policy is not to expose internal grant/purchasing fields publicly without permission. Keep grants in graph/internal layers only.
  3. Overstating inferred publication links. PUB-009→CSH-016 is include_in_graph: false (method alignment). If surfaced, it must read as supporting/related, not as a facility acknowledgment.
  4. Hand-maintained link lists drift. ## Linked Relationship IDs are prose, not generated; manual edits can desync from people.yaml / *_links.yaml. Any refactor should keep them consistent with the YAML evidence.
  5. Edit the right layer. These changes belong in resources/CSH/*.md source (and possibly people.yaml evidence), never in docs/ (generated mirror). Rebuild via sync_docs.sh + mkdocs build --strict.
  6. Equipment "Last verified" mismatch. Equipment blocks show a hardcoded May 28, 2026 vs the page's May 30, 2026; not a Grice-cluster-specific issue but visible here.

Ordered by value-to-risk. No action taken yet — awaiting review.

# Action Risk Evidence source
1 Add CSH-002 to CSH-017 Linked Relationship IDs Low people.yaml PER-074 director edges + catalog
2 Add PUB-013 DOI reference to CSH-017 (direct, produced_by) Low publications.yaml PUB-013
3 Convert PUB-009 on CSH-016 into a DOI link, framed as supporting Low publications.yaml PUB-009
4 Inline-link in-prose resource IDs on all three pages Very low targets already exist
5 Cite Grice–Karver coauthorship in CSH-016 caspase note Low people.yaml PER-100→CSH-017
6 (Optional) Surface person evidence_text in a review-facing view Low people.yaml
7 (Defer) Evaluate CSH-016↔CSH-018 only if direct evidence is found Do not act now none currently
8 (Defer) Keep grants internal; do not surface on pages n/a policy

Suggested vehicle: package actions 1–5 as a single source packet (e.g. sources/inbox/2026-06-13-grice-lab-pilot/), produce an evidence-backed diff, review via Git, then sync_docs.sh + strict build before any move toward main.

Bottom line

The three pages are solid and honestly scoped. The improvement opportunity is almost entirely surfacing and linking knowledge the repository already holds — one unlinked director relationship (CSH-002), two publications that exist in YAML but are invisible or unresolvable on the pages, several in-prose IDs that should be hyperlinks, and one coauthorship citation. These are low-risk, evidence-backed refactors. The one symmetry temptation (Grice↔Southern) should be resisted absent direct evidence.