You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Historical CAS-RN cohort: decide identities for 6 MediaDive cas: nodes and ~60 IDs now on ungrounded placeholders #286
Recovered a concrete historical cohort for this issue; chemical resolution is still pending.
The first-column CAS-RN: query in the pre-issue Makefile, applied to all eight committed source node tables at that revision, yields 234 distinct literal IDs: 42 BacDive and 192 MediaDive. An independent streaming comparison finds exact set equality with all 234 CAS node IDs in the official March 7, 2025 core archive, not merely an equal count. Original malformed spellings and a trailing nonbreaking space are preserved.
This strongly corroborates the historical cohort but does not prove it is the exact February 11 graph originally examined. The downloaded asset has observed SHA-256 a7eebfc255ed39435cdd2f28754b5bd6b063b3c58fced02477ba687c2fd3e953; GitHub supplies no upstream content digest for it. The sorted 234-ID list has SHA-256 5bb6fba8f8fea8a875e2e3e5a1be0e47664685d1705c40eebf50754b24a8d490.
A finite diagnostic against the reviewed current unified candidate (09c44642ab13b234e89ce113f7510aa9efb56969e4b539cb7843b43dcb425ba7) and unchanged supported MIM table accounts for every historical ID:
Current representation
IDs
Explicit CAS-to-ChEBI table claim
5
Exact MIM-to-CAS primary claim, not ChEBI equivalence
3
Existing finite CAS scope policy
1
CAS endpoint only in lexical rows
8
Runtime name candidate only
116
No endpoint, scope-policy or runtime name candidate in this bounded diagnostic
101
Total
234
These are representation buckets, not resolved-identity counts. Name matches, ontology xrefs and syntactically valid CAS identifiers do not by themselves establish exact material identity. Five original literals have shape/whitespace/checksum findings; none was silently repaired. Current raw-source uses of the strongest label/identifier discrepancies are being traced before proposing any additional runtime change.
Project-relative retained evidence is under data/issue1224-quarantine-20260929.xjuS9H/:
issue286-independent-core.BXSXby/README.md: full original archive/source evidence and exact set comparison.
issue286-current-mapping.WdPFGy/README.md: diagnostic boundaries and mapping pins; SHA-256 b588d5fc2877772b2872073627d022f3d0c86a31d2234cbcf908efd8ff8e8645.
issue286-current-mapping.WdPFGy/queue.json: all 234 IDs and 516 original source/archive rows; SHA-256 95362365feb46fb29bf1b1960db4624f99174e0a9be20b7681f12b0c113cdee0.
Separately, the narrow Potato/extract correction in #1239 has passed real isolated MediaDive replay and full-row comparison; revised full repository gates and CI are running. The final all-source rebuild and merged-candidate validation remain pending. This comment does not close #286 or claim release acceptance.
Fresh candidate reconciliation is complete for the bounded recovered historical cohort. Keeping #286 open for the separate unresolved chemical-identity/assay curation; this does not claim all 234 identities are resolved.
The all15 rebuild and isolated candidate merge are complete at landed 1408e7099d039026d7611c240938d8e177753406; candidate archive SHA-256 158e7131025c7941ae98d70af5ca14d6efb69857ed019f7f61fa9b44279b2345.
Actual CAS checking preserves six current CAS endpoints and 110 full assertions. Five historical IDs match only through explicitly recorded CAS-RN: → cas: prefix normalization (108 assertions); the sixth is outside the recovered cohort. The other 229 historical endpoint absences do not prove identity resolution or loss of observations.
All 5,473 saved MediaDive occurrences retain literal raw records, typed compound/solution IDs, quantities and non-target context through the candidate merge. They cover 191 historical MediaDive IDs; the saved evidence has no occurrence for CAS-RN:94-75-7. This is a bounded evidence gap, not exhaustive absence from contemporary data.
Exactly 29 target changes are separately enumerated: 28 material-quarantine-audited changes and one existing fresh mapping claim for tetramethyl ammonium chloride. None is promoted to new identity approval by this join. All nine historical ambiguous display groups remain explicit.
The 42-ID/72-row saved BacDive diagnostic remains assay-context evidence. Its one previously explicit EC/substrate route is joined to the corrected EC:3.2.1.20 → CHEBI:91122 assertion and exact original-row correction audit; candidate assay similarities are not new CAS equivalences.
All 234 original literals, 516 original physical rows, original mapping epochs and false identity-approval flags remain retained. The earlier 116 name-only / 101 no-candidate buckets are historical diagnostic counts, not a newly verified current backlog.
The independently reviewed actual whole-source projection preserved all 113,918 MediaDive assertions and all 16 finalized BacDive EC/substrate assertions, with zero missing/extra/duplicate owned assertions; #1265/#1273 are closed for those bounded checks. That does not close this broader curation issue or authorize release.
Remaining, within the identity and assay curation kept open on 2026-09-30 (counts from the 2026-10-13 build):
Six MediaDive ingredients are still cas: nodes, on 110 edges (Bacto peptone 103): KH2PO3, Bacto peptone, TAPSO, TitaniumIII chloride, CrKSO42 x 12 H2O, Cheese whey. The reviewed refresh preserved four of these (docs/reviews/mediadive-potato-scope-20260929/PROMOTION.md:65-69); TAPSO, outside the cohort, has a reviewed CHEBI:84015 candidate (mappings/culturebotai_reviewed_ingredients.tsv:789 at 1acb8b6); peptone and whey are mixtures.
By exact name or synonym, 60 of the 192 historical MediaDive IDs now match only mediadive.ingredient placeholders: 67 nodes on 662 edges, none with an xref or same_as (e.g. 'LaCl3 x 7 H2O'). MediaDive's ingredient table (fetched 2026-10-15) gives all 67 a CAS and none a ChEBI ID, so Download MediaDive's bulk ingredients table (group, native ChEBI and CAS ids) #383's bulk download alone would restore only the CAS.
7 historical MediaDive IDs match no current name or synonym (e.g. 'MOPS buffer'); trace them.
Decide whether the 42 BacDive assay-substrate IDs need more than the retained assay context (by exact name, 12 match a CHEBI node in the BacDive output and 30 match nothing).
Acceptance: each ID in items 1-3 has a recorded decision (a reviewed identity or an explicit keep), and after the next MediaDive rebuild the cohort's cas: and placeholder counts in data/transformed/mediadive/nodes.tsv match those decisions.
changed the title [-]234 unmapped CAS-RN in current merge[/-][+]Historical CAS-RN cohort: decide identities for 6 MediaDive cas: nodes and ~60 IDs now on ungrounded placeholders[/+]on Oct 16, 2026
Activity
realmarcin commentedon Sep 29, 2026
Recovered a concrete historical cohort for this issue; chemical resolution is still pending.
The first-column
CAS-RN:query in the pre-issue Makefile, applied to all eight committed source node tables at that revision, yields 234 distinct literal IDs: 42 BacDive and 192 MediaDive. An independent streaming comparison finds exact set equality with all 234 CAS node IDs in the official March 7, 2025 core archive, not merely an equal count. Original malformed spellings and a trailing nonbreaking space are preserved.This strongly corroborates the historical cohort but does not prove it is the exact February 11 graph originally examined. The downloaded asset has observed SHA-256
a7eebfc255ed39435cdd2f28754b5bd6b063b3c58fced02477ba687c2fd3e953; GitHub supplies no upstream content digest for it. The sorted 234-ID list has SHA-2565bb6fba8f8fea8a875e2e3e5a1be0e47664685d1705c40eebf50754b24a8d490.A finite diagnostic against the reviewed current unified candidate (
09c44642ab13b234e89ce113f7510aa9efb56969e4b539cb7843b43dcb425ba7) and unchanged supported MIM table accounts for every historical ID:These are representation buckets, not resolved-identity counts. Name matches, ontology xrefs and syntactically valid CAS identifiers do not by themselves establish exact material identity. Five original literals have shape/whitespace/checksum findings; none was silently repaired. Current raw-source uses of the strongest label/identifier discrepancies are being traced before proposing any additional runtime change.
Project-relative retained evidence is under
data/issue1224-quarantine-20260929.xjuS9H/:issue286-independent-core.BXSXby/README.md: full original archive/source evidence and exact set comparison.issue286-current-mapping.WdPFGy/README.md: diagnostic boundaries and mapping pins; SHA-256b588d5fc2877772b2872073627d022f3d0c86a31d2234cbcf908efd8ff8e8645.issue286-current-mapping.WdPFGy/queue.json: all 234 IDs and 516 original source/archive rows; SHA-25695362365feb46fb29bf1b1960db4624f99174e0a9be20b7681f12b0c113cdee0.Separately, the narrow Potato/extract correction in #1239 has passed real isolated MediaDive replay and full-row comparison; revised full repository gates and CI are running. The final all-source rebuild and merged-candidate validation remain pending. This comment does not close #286 or claim release acceptance.
realmarcin commentedon Sep 30, 2026
Fresh candidate reconciliation is complete for the bounded recovered historical cohort. Keeping #286 open for the separate unresolved chemical-identity/assay curation; this does not claim all 234 identities are resolved.
The all15 rebuild and isolated candidate merge are complete at landed
1408e7099d039026d7611c240938d8e177753406; candidate archive SHA-256158e7131025c7941ae98d70af5ca14d6efb69857ed019f7f61fa9b44279b2345.CAS-RN:→cas:prefix normalization (108 assertions); the sixth is outside the recovered cohort. The other 229 historical endpoint absences do not prove identity resolution or loss of observations.CAS-RN:94-75-7. This is a bounded evidence gap, not exhaustive absence from contemporary data.EC:3.2.1.20 → CHEBI:91122assertion and exact original-row correction audit; candidate assay similarities are not new CAS equivalences.The independently reviewed actual whole-source projection preserved all 113,918 MediaDive assertions and all 16 finalized BacDive EC/substrate assertions, with zero missing/extra/duplicate owned assertions; #1265/#1273 are closed for those bounded checks. That does not close this broader curation issue or authorize release.
Retained project-relative evidence:
data/issue1224-quarantine-20260929.xjuS9H/fresh-source-acceptance-1408/issue286-context-join-02/README.md:4a7f7955740943ffb2aa1caef1a882e0f8effbec00558b9b15ed74a22afeb258.summary.json:25ed832df7098bf37ec03249ddd4e494c681926ab230496dde4ef9ceaf6ea0dc.backbone-with-context.jsonl:182510290b7c05c7c7994cca5c299665220b270c204c291b9d86b0ae2f257dea.The canonical archive and user statistics were not replaced. Full candidate review/retention and publication remain separate.
realmarcin commentedon Oct 16, 2026
Narrowing this to what remains (triage of 2026-10-15).
Done:
CAS-RN:prefix.Remaining, within the identity and assay curation kept open on 2026-09-30 (counts from the 2026-10-13 build):
cas:nodes, on 110 edges (Bacto peptone 103): KH2PO3, Bacto peptone, TAPSO, TitaniumIII chloride, CrKSO42 x 12 H2O, Cheese whey. The reviewed refresh preserved four of these (docs/reviews/mediadive-potato-scope-20260929/PROMOTION.md:65-69); TAPSO, outside the cohort, has a reviewed CHEBI:84015 candidate (mappings/culturebotai_reviewed_ingredients.tsv:789at 1acb8b6); peptone and whey are mixtures.mediadive.ingredientplaceholders: 67 nodes on 662 edges, none with an xref or same_as (e.g. 'LaCl3 x 7 H2O'). MediaDive's ingredient table (fetched 2026-10-15) gives all 67 a CAS and none a ChEBI ID, so Download MediaDive's bulk ingredients table (group, native ChEBI and CAS ids) #383's bulk download alone would restore only the CAS.Acceptance: each ID in items 1-3 has a recorded decision (a reviewed identity or an explicit keep), and after the next MediaDive rebuild the cohort's
cas:and placeholder counts indata/transformed/mediadive/nodes.tsvmatch those decisions.🤖 Generated with Claude Code
[-]234 unmapped CAS-RN in current merge[/-][+]Historical CAS-RN cohort: decide identities for 6 MediaDive cas: nodes and ~60 IDs now on ungrounded placeholders[/+]