MangroveNP v0.6

帮助 / Help

中文

1. 如何检索

首页文献证据浏览

• 在检索框上方选择检索方式：按化合物名称、按物种、按文献（如 DOI 文本），或按结构（见第 2 节）。
• 可按物种、证据类型（原始全文 / 原始摘要 / 原始补充数据，以及 MSI Level 1/2 分析注释）等条件筛选。

公开结构库

• 检索类型：综合检索、化合物名称、MangroveNP 编号、InChIKey、分子式，或源 SMILES（精确字符串）。综合检索会在名称、编号、InChIKey、SMILES、分子式、外部标识符、物种和 DOI 中查找，多个词须同时出现。
• 筛选：按结构来源（structure_from）、按物种、按是否有结构。结构来源只有三个选项：COCONUT、LOTUS、OPSIN；有多个来源的记录（如同时来自 COCONUT 与 LOTUS）在其每个来源下都会出现。交叉引用数据库（also_listed_in，如 NPAtlas、PubChem）不能用于筛选，只在记录详情页显示。
• 按来源生物类型筛选：检索界面只显示本版本有记录的四种类型：植物、真菌、细菌、原生生物。一条记录可能属于多种类型，因此各类型的记录数相加会超过记录总数。
• 排序：按名称或按结构来源（按 structure_from 的第一项，相同时按记录号的数值）；每页 25 / 50 / 100 条。
• 筛选结果可导出为 CSV 或 JSON；导出包含全部匹配记录，不只当前页。

按 NPClassifier 类别浏览

• 可按 NPClassifier 类别（pathway / superclass / class）浏览。
• 有多个类别的记录会出现在其每一个类别下。最高一级没有类别的记录归入「未分类 / Unclassified」；已有上一级类别、但下一级没有类别的记录归入「未细分 / Not further classified」。
• 因此各类别的计数相加会大于记录总数：有多个类别的记录在其每个类别下各计一次。

2. 结构检索

• 相似性检索：输入查询结构，返回 Tanimoto 相似度 ≥ 0.6（含 0.6）的结构，按 Morgan 指纹（半径 2，2,048 位）的 Tanimoto 相似度排序。阈值默认 0.6，可在 0.30–0.90 之间以 0.05 为步长调整；超出范围的阈值按边界值处理。
• 子结构检索：输入 SMILES 或 SMARTS，返回含该子结构的记录。
• 计算均用 RDKit.js 2026.9.1 在浏览器中完成，结构查询不发送到服务器。
• 有 4 个 InChIKey 各对应多条记录，这些记录的 SMILES 是不同的互变异构体写法。结构检索索引中，每个这样的 InChIKey 只使用记录号最小的那条记录的结构；记录详情页仍显示各记录自己的结构。
• 查询结构在文本框中以 SMILES 输入；子结构检索中勾选「SMARTS」时按 SMARTS 解析。精确检索还接受完整 InChIKey 或其第一段（前 14 个字符）。不提供结构绘制工具。
• 结构检索结果默认每页 20 条；每个命中为一个 InChIKey，列出具有该 InChIKey 的全部记录。相似性检索结果按 Tanimoto 相似度降序、再按记录号排列；子结构与精确检索结果按记录号排列。「导出全部」导出全部命中，而不只是当前页。

3. 如何阅读一条记录

• 结构：结构图、SMILES、InChIKey 与分子式。结构来源（COCONUT、LOTUS 或 OPSIN 名称转结构）逐条注明；同一结构也见于的交叉引用数据库列在详情页，仅作交叉引用。
• 匹配置信度（高 / 中 / 低）：表示名称与结构匹配的把握程度，不是逐条人工审核结论。
• 原文分子式比对：一致 / 仅电荷或盐形式不同 / 不一致 / 无原文分子式。"不一致"本身不等于结构错误。
• 结构提示标记：少数记录带 ambiguous_name 或 demoted_annotation，表示原文不足以唯一确定结构，请结合原始文献使用。
• 候选标识符：CAS 号均为校验位通过的候选号，并附 strict_class 与来源；PubChem CID、Wikidata QID、ChEBI ID 由自动交叉匹配生成。InChIKey 相同的记录显示相同的标识符。
• 文献证据：生物来源、来源类别（植物 / 真菌 / 细菌 / 原生生物）、部位（如有）和原始文献链接。

各字段的准确含义见"数据字典"页。

4. 下载与校验

• 下载文件位于 data/ 目录：公开结构层（mangrovenp-v0.6-public.json 与 mangrovenp-v0.6-public.csv）、结构文件（mangrovenp-v0.6-structures.sdf 与 mangrovenp-v0.6-structures.smi）、文献证据层（mangrovenp-v0.6-literature.json）、描述符表与 NPClassifier 分类表（TSV）、每条记录的 SVG 结构图、外部数据库名录（resources.json：本站引用的外部数据库及其各自许可，仅作说明），以及校验文件 SHA256SUMS。均以 CC BY 4.0 发布；resources.json 所列各数据库的内容不随本站分发，其许可以各库自己的声明为准。
• SHA256SUMS 覆盖 data/ 目录下除其自身以外的全部文件：各下载文件、每条记录的 SVG 结构图（data/svg/）、网站数据文件（data/db/）与结构检索索引（data/struct/）等；其中的路径相对于 data/ 目录。

Linux / macOS（在下载文件所在目录中运行；未下载全部 SVG 时加 --ignore-missing）：

sha256sum -c --ignore-missing SHA256SUMS

Windows PowerShell（将输出与 SHA256SUMS 中同名文件的一行比对，大小写不影响）：

Get-FileHash .\mangrovenp-v0.6-public.json -Algorithm SHA256

5. 如何引用

请见"如何引用"页，并注明所用版本号。

6. 报告错误或提出更正

纠错渠道即将公布；告知时请注明记录号、InChIKey 与问题说明。勘误记录列于"勘误"页。

7. 常见问题

• 为什么找不到 v0.5 中的某条记录？ 可能属于 v0.5 冻结后发现的勘误记录，已从 v0.6 公开层移除，见"勘误"页。
• CAS 号可以直接引用吗？ 本库 CAS 号均为候选号，不是 CAS 官方授权数据，使用前请核对权威来源。
• 本库可用于临床或用药决策吗？ 不可以。本库为研究索引，详见"免责声明"页。
• 数据多久更新一次？ 不定期更新，每次发布附变更记录。

English

1. How to search

Literature evidence explorer (home page)

• Above the search box, choose how to search: by compound name, by species, by publication (for example DOI text), or by structure (see section 2).
• Results can be filtered by species and by evidence type (primary full text / primary abstract / primary supplementary data, and MSI Level 1/2 analytical annotations), among other options.

Public structure library

• Search types: any field, compound name, MangroveNP record ID, InChIKey, molecular formula, or source SMILES (exact string). "Any field" looks in names, record IDs, InChIKeys, SMILES, formulas, external identifiers, species and DOIs; all words must match.
• Filters: structure source (structure_from), species, and whether a structure is available. The structure source has exactly three options: COCONUT, LOTUS and OPSIN; a record with several sources (for example both COCONUT and LOTUS) appears under each of them. Cross-reference databases (also_listed_in, such as NPAtlas or PubChem) cannot be used as filters and are shown on the record detail page only.
• Source organism type filter: the interface shows only the four types that have records in this release: plant, fungus, bacterium and protist. A record may belong to more than one type, so the per-type counts add up to more than the total number of records.
• Sort by name or by structure source (by the first item of structure_from, ties broken by the numeric record number); 25 / 50 / 100 records per page.
• Filtered results can be exported as CSV or JSON; the export covers every matching record, not only the current page.

Browsing by NPClassifier class

• Records can be browsed by NPClassifier class (pathway / superclass / class).
• A record with several labels appears under each of its labels. Records without a label at the top level are shown under 「未分类 / Unclassified」; records that have a label at the level above but none at the next level are shown under 「未细分 / Not further classified」.
• Counts per category therefore add up to more than the total, because a record with several labels is counted under each of its categories.

2. Structure search

• Similarity search: enter a query structure; structures with Tanimoto ≥ 0.6 (0.6 included) are returned, ranked by Tanimoto similarity of Morgan fingerprints (radius 2, 2,048 bits). The threshold defaults to 0.6 and can be set from 0.30 to 0.90 in steps of 0.05; a threshold outside this range is treated as the nearest boundary value.
• Substructure search: enter SMILES or SMARTS; records containing the substructure are returned.
• Both are computed in your browser with RDKit.js 2026.9.1; structure queries are not sent to the server.
• For 4 InChIKeys, the records sharing the InChIKey carry different tautomeric SMILES. The structure-search index uses, for each such InChIKey, the structure of the record with the smallest record number; each record's detail page still shows that record's own structure.
• Enter the query structure as SMILES in the text box; in substructure search, tick "SMARTS" to have it read as SMARTS. Exact search also accepts a full InChIKey or its first block (the first 14 characters). There is no structure drawing tool.
• Structure-search results are shown 20 per page by default; each hit is one InChIKey and lists all records with that InChIKey. Similarity hits are sorted by Tanimoto similarity (descending), then by record number; substructure and exact hits are sorted by record number. "Export all" exports every hit, not only the current page.

3. How to read a record

• Structure: structure image, SMILES, InChIKey and molecular formula. The structure source (COCONUT, LOTUS or OPSIN name-to-structure conversion) is given for each record; cross-reference databases in which the same structure is also listed are shown on the detail page, for cross-reference only.
• Match confidence (high / medium / low): how confidently the name was matched to the structure; it is not the result of an individual manual review.
• Source-formula comparison: consistent / charge or salt form only / inconsistent / no comparable source formula. "Inconsistent" alone does not mean the structure is wrong.
• Structure caution flags: a few records carry ambiguous_name or demoted_annotation, meaning the original publication is not sufficient to determine the structure uniquely; please consult the original publication.
• Candidate identifiers: CAS numbers are check-digit-passing candidates only, each with strict_class and source; PubChem CIDs, Wikidata QIDs and ChEBI IDs come from automated cross-matching. Records sharing an InChIKey show the same identifiers.
• Literature evidence: source organism, organism category (plant / fungus / bacterium / protist), plant part where reported, and a link to the original publication.

For the exact meaning of each field, see the Data dictionary page.

4. Downloads and checksums

• Download files are in the data/ directory: the public structure layer (mangrovenp-v0.6-public.json and mangrovenp-v0.6-public.csv), the structure files (mangrovenp-v0.6-structures.sdf and mangrovenp-v0.6-structures.smi), the literature evidence layer (mangrovenp-v0.6-literature.json), the descriptor and NPClassifier tables (TSV), one SVG structure image per record, the directory of external databases (resources.json: the external databases referred to by this site with their own licences, for information only), and the checksum file SHA256SUMS. All are released under CC BY 4.0; the contents of the databases listed in resources.json are not redistributed by this site, and their licences are as stated by each database.
• SHA256SUMS covers every file under the data/ directory except itself: the download files, the per-record SVG structure images (data/svg/), the site data files (data/db/) and the structure-search index (data/struct/), among others; paths in it are relative to the data/ directory.

Linux / macOS (run in the directory that holds the downloaded files; add --ignore-missing if you did not download every SVG):

sha256sum -c --ignore-missing SHA256SUMS

Windows PowerShell (compare the output with the line for the same file in SHA256SUMS; letter case does not matter):

Get-FileHash .\mangrovenp-v0.6-public.json -Algorithm SHA256

5. How to cite

See the How to cite page, and state the release you used.

6. Report an error or suggest a correction

The correction channel is coming soon; when telling us, please give the record ID, InChIKey and a description. Erratum records are listed on the Errata page.

7. Frequently asked questions

• Why can't I find a record that was in v0.5? It may be one of the erratum records found after the v0.5 freeze, which were removed from the v0.6 public layer; see the Errata page.
• Can I cite the CAS numbers directly? All CAS numbers here are candidates, not an official CAS Registry product; please confirm against authoritative sources before use.
• Can MangroveNP be used for clinical or treatment decisions? No. It is a research index; see the Disclaimer page.
• How often is the data updated? Updated irregularly; each release comes with a changelog.

数据许可：CC BY 4.0 · Data licence: CC BY 4.0 — 许可与引用 · License & citation

返回首页 · Back to home