Stopped without a submission because faithful, complete curation requires inaccessible supporting information and resolution of a material source conflict. Live wishlist item 126570036 was Available with no assignees. Read pinned skill and verifier contract vct_4t6ri8zibj; reference card SHA-256 matched f71e58f620f577e1d08cbd1915249fb01d40e3d2d67addb983b759242055fd71.
Primary CC-BY-4.0 data at Zenodo 7106075 contain 454 experimental optimization rows (round counts 154/72/72/72/84) plus Figure4A's 80 test-set rows (20 products x four conditions). Preserve rows separately, multiply fractional optimization yields by100, and leave test yields already in percent. Figure4B is only a row permutation of the Figure4A yield matrix; Figure4C/D are derivative statistics. Figure4E/F/G report additional UV area ratios to attach to test reactions. Exclude literature-mined Figure3C data and ML predictions. Main article explicitly identifies extra SI time-extension and platform-control experiments, so534cannot be claimed complete.
The authors' Zenodo7099435 archived code contains their draft ORD template and534-row source CSV. It encodes0.3mmol MIDA boronate whereas corresponding patent EP4608988A1 GP3 gives0.12mmol, both with0.1mmol halide. Do not silently choose. Publisher exact SI URL returned403; legitimate author lab pages supplied the main article only, PMC/Figshare checks were empty, and the author's Illinois thesis normal retrieval also returned403. No paywall or bot checks were bypassed.
Structure audit matters beyond RDKit parsing:169optimization rows omit a carbonyl in the stated MIDA group; Figure4AProduct29 reactant/product encode a radical amine instead of the pictured nitrile; archived ORD precursor SMILES for P(tBu)3PdG4,PCy3PdG4,and Pd(dppf)Cl2 are chemically defective. Use explicit source-backed corrections. Preserve phenanthrene as a pre-reaction INTERNAL_STANDARD input because the experimental method preloads it.
No dataset, validator output, artifact upload, TfP submission or upstream ORD PR was produced. Resuming requires exact paper SI or author clarification, full supplementary coverage audit, source-license review, documented structure corrections, actual ord-schema0.8.3validation and independent curation review. Work used GPT-6Astra assistance, Python/pandas/RDKit/pypdf and direct legitimate source retrieval. This attempt is not counted toward the user's five complete problems; the broader goal continues with an accessible replacement.
The authors' archived ORD template at Zenodo 7099435 uses 0.3 mmol MIDA boronate, but their corresponding patent GP3 uses 0.12 mmol for the same 0.1 mmol halide scale. Exact paper SI is required to resolve this; a silent choice would risk fabricated conditions. Publisher SI returned 403, author laboratory pages provide the main article only, and normal checks of Figshare/PMC/author thesis did not provide the SI. No access controls were bypassed. Main article also identifies additional SI time-extension/platform controls, preventing a complete scope claim from the 454+80 core rows. Chemical audit found 169 optimization rows missing one MIDA carbonyl in otherwise parseable SMILES, test Product29 encoding a radical amine instead of the pictured nitrile, and misencoded G4 precatalysts in the authors' draft. These require documented corrections. Findings and exact source paths are checkpointed for future continuation. No dataset, validator result, artifact upload, submission, or upstream ORD PR has been represented as completed.
RDKit parses all core source SMILES, but this alone is insufficient. Four distinct optimization boronate strings (169 rows) omit one carbonyl oxygen in the MIDA protecting group, encoding an aminoethanol-derived chelate instead of the stated MIDA boronate. The test-set files also need name/structure comparison beyond parsing. Corrections will retain original strings in a source audit and require agreement with the paper's depicted/named compounds. Discovered the authors' Zenodo 7099435 linked code release includes open_reaction_database_files/reaction_template.pbtxt and dataset_withtestset.csv/.pbtxt; inspecting these as corroborating primary sources, not treating author-generated records as validated.
Downloaded the CC-BY-4.0 Zenodo figure data and verified 454 experimental optimization rows: rounds 1–5 contain 154, 72, 72, 72 and 84 rows. Only first-round rows carry A/B experiment IDs; later rows must use original row indices for traceability. Figure 4A.csv contains another 80 experimental test-set entries (20 products × four conditions), including structures and percent yields. Figure 4B.xlsx is the same yield matrix in a different product order, not additional reactions. The author-hosted article on Schroeder's Princeton site identifies further SI time-course and byproduct experiments, so 534 rows alone cannot yet be claimed complete. Direct publisher SI access returned HTTP 403; checking legitimate author/repository alternatives, without bypassing access controls. The authors' corresponding patent EP4608988A1 reproduces experimental methods and may provide corroboration, but paper-specific SI coverage remains under audit.
The live GitHub wishlist item 126570036 is Available with no assignees. Read the pinned curation skill and contract and verified the reference card SHA-256 f71e58f620f577e1d08cbd1915249fb01d40e3d2d67addb983b759242055fd71. Plan: retrieve authors' CC-BY-4.0 Zenodo data, retain all 454 experimental optimization rows as separate replicate Reactions, inspect Figure 4 data for additional reported experiments, and obtain SI for quantities, time, solvent ratio and analytical method. Exclude literature-mined rows. Build reproducible ORD script, validate with ord-schema 0.8.3, and audit chemical structures and full source coverage before independent review. This is user-initiated work with GPT-6 Astra assistance; no upstream ORD submission is being made.