MangroveNP v0.6
更新日志 / Changelog
中文
v0.6 —— 公开版（准备中；发布日期即将公布）
- 下载文件：mangrovenp-v0.6-public.json，许可 CC BY 4.0。规模：6,234 条记录 / 5,773 个唯一 InChIKey。
- 公开范围：仅提供经整理的公开层；候选/交叉对照层不公开；公开版无登录或写入接口。
- 文献证据层：data/mangrovenp-v0.6-literature.json，6,234 条化合物条目（每条公开记录一条）、共 9,823 条逐条证据，来自 901 篇文献；仅保留被公开化合物记录引用的文献来源。
- 新增：公开外部标识符：CAS 候选号（校验位 + strict_class + 来源）、PubChem CID（cid_primary / cid_all_exact_ik）、Wikidata QID、ChEBI ID。覆盖数（截至 2026-10-19，Wikidata 待回填；分母为 5,773 个公开 InChIKey）：候选 CAS 3,055（52.92%）/ 候选 CAS（主）2,419（41.90%）/ CID 5,504（95.34%）/ QID 4,626（80.13%）/ ChEBI 3,389（58.70%）。
- 新增：计算描述符（RDKit）：可下载表格 data/mangrovenp-v0.6-descriptors.tsv，每个 InChIKey 一行，共 5,773 行。
- 新增：NPClassifier 分类（pathway / superclass / class）：可下载表格 data/mangrovenp-v0.6-npclassifier.tsv，每个 InChIKey 一行，共 5,773 行。
- 勘误：v0.5 冻结后发现的 15 条勘误记录已从公开层移除（其中 3 条为后续发现，结构与原文不符；另有 1 条随后撤回：原文为 GC-MS 谱库匹配，所列名称是氘代标记物，不是从样品中分离鉴定的化合物（详见"勘误"页的勘误表）；−15 条记录 / −10 个 InChIKey；6,249 / 5,783 → 6,234 / 5,773）。
- 更正：1 条记录的原文分子式录入更正（结构不变）。
- 下载地址：即将公布 · Zenodo（DOI 即将公布）
v0.5 —— 冻结版（论文所用；2026-10-14 构建）
- 文献整理化合物记录 10,614 条；保留结构 7,444 条（6,826 个唯一 InChIKey）。
- v0.5 开放许可公开层（冻结）6,249 条记录 / 5,783 个唯一 InChIKey。
- 冻结数据：v0.5 冻结版。
- 冻结后勘误：截至 v0.6，15 条公开层勘误记录已列出，延至下一版本修正；v0.5 主要计数不变。
- Zenodo（DOI 即将公布）
English
v0.6 — public release (in preparation; release date coming soon)
- Download file: mangrovenp-v0.6-public.json, licensed CC BY 4.0. Size: 6,234 records / 5,773 unique InChIKeys.
- Public scope: curated public layer only; the candidate/crosswalk layer is not public; the public build has no login or write endpoints.
- Literature evidence layer: data/mangrovenp-v0.6-literature.json, 6,234 compound entries (one per public record) with 9,823 evidence items from 901 sources; only literature sources cited by a public compound record are kept.
- Added: public external identifiers: CAS candidates (check digit + strict_class + source), PubChem CIDs (cid_primary / cid_all_exact_ik), Wikidata QIDs, ChEBI IDs. Coverage (as of 2026-10-19; Wikidata back-fill pending; denominator 5,773 public InChIKeys): candidate CAS 3,055 (52.92%) / primary candidate CAS 2,419 (41.90%) / CID 5,504 (95.34%) / QID 4,626 (80.13%) / ChEBI 3,389 (58.70%).
- Added: computed descriptors (RDKit): downloadable table data/mangrovenp-v0.6-descriptors.tsv, one row per InChIKey, 5,773 rows.
- Added: NPClassifier classes (pathway / superclass / class): downloadable table data/mangrovenp-v0.6-npclassifier.tsv, one row per InChIKey, 5,773 rows.
- Errata: the 15 erratum records found after the v0.5 freeze have been removed from the public layer (including three further records whose structure does not match the source, and one further record withdrawn subsequently: the source reports a GC-MS library match, and the listed name is a deuterium-labelled compound, not a compound isolated and identified from the sample (see the errata table on the Errata page); −15 records / −10 InChIKeys; 6,249 / 5,783 → 6,234 / 5,773).
- Corrected: transcription of the source molecular formula for 1 record (structure unchanged).
- Download: Coming soon · Zenodo (DOI coming soon)
v0.5 — frozen release (used in the paper; built 2026-10-14)
- 10,614 literature-compiled compound records; 7,444 structure-resolved records (6,826 unique InChIKeys).
- v0.5 openly licensed public layer (frozen): 6,249 records / 5,783 unique InChIKeys.
- Freeze file: the v0.5 frozen release.
- Post-freeze errata: as of v0.6, fifteen public-layer erratum records documented and deferred to the next release; v0.5 headline counts unchanged.
- Zenodo (DOI coming soon)
数据许可：CC BY 4.0 · Data licence: CC BY 4.0 — 许可与引用 · License & citation