Can Immune

io.github.sanjaysgk由 GitHub 帳號 sanjaysgk 發布v1.0.0更新於 Oct 24, 2026

Query CAN-IMMUNE: cancer neoantigen mutations, peptides, cell lines, MHC-I binding. Read-only.

已驗證Streamable HTTP可網頁執行知識與記憶資料與分析

概覽

AI 產生的概覽

一個針對 CAN-IMMUNE 癌症新抗原資料庫的遠端唯讀 MCP 伺服器,讓助理查詢突變、胜肽、細胞株、組織及快取的 MHC-I 結合預測。

可以做什麼
取得資料庫中突變、基因、胜肽、細胞株、組織與癌症類型的總量統計及資料來源。依名稱搜尋基因與細胞株並取得完整檔案,包括高頻復發突變與突變最多的基因。依突變負荷排序列出癌症類型與原發組織。列出某個基因的逐筆突變,含胺基酸改變、CDS 改變、樣本與突變胜肽。查詢某胜肽與 HLA 對偶基因的快取 MHC-I 結合預測。透過通用連接器介面為 ChatGPT 類用戶端提供搜尋與記錄取得。
適用情境
適合回答關於新抗原突變負荷、突變胜肽、細胞株模型與快取 MHC-I 結合預測的癌症基因體學或免疫學研究問題,用自然語言即可,無須自行撰寫資料庫查詢。
執行需求
遠端 Streamable HTTP 端點,無須本機執行環境或安裝套件。匿名使用有較低速率限制;可選 Google 登入(OAuth 2.1)可提高限額,學術信箱可獲得最高等級。需要網路存取。
安裝前請注意
雖為唯讀,但每次查詢都會傳送到莫納什大學的第三方研究服務,搜尋詞會傳給該營運方。MHC-I 結合結果僅為快取預測,不會執行新的預測。資料來自 COSMIC、DepMap/CCLE 與 PubMed,用於研究或臨床前應與原始來源核對。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 Can Immune,將其新增到工作區。
  2. 為需要使用其工具的對話啟用該服務。

Web executable,透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。

交給代理程式安裝

把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式,或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南,先向你展示發布者、端點和登入方式,經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。

閱讀 https://sourceweft.com/mcp/io.github.sanjaysgk%2Fcan-immune/install.md,並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前,先告訴我它的發布者、端點和登入方式,等我確認。不要把 API 金鑰寫進命令或我會提交的檔案;問我金鑰應從哪裡取得。

檢視代理程式所依據的安裝指南

其他 MCP 客戶端

把它新增到你客戶端的 mcpServers 設定中。

{
  "mcpServers": {
    "can-immune": {
      "type": "http",
      "url": "https://canelib.erc.monash.edu/mcp"
    }
  }
}

README

CAN-IMMUNE MCP Server

Remote, read-only Model Context Protocol server for CAN-IMMUNE — a cancer neoantigen platform. Lets any AI agent (Claude, ChatGPT, Cursor, ...) query the database in plain language.

Connect

Add the endpoint as a custom connector, or:

claude mcp add --transport http canelib https://canelib.erc.monash.edu/mcp

Tools (read-only)

database_overview, search / fetch, search_genes, get_gene, search_cell_lines, get_cell_line, top_genes_in_cell_line, get_gene_mutations, list_tissues, list_cancer_types, get_mhc_binding.

Auth & rate limits

Works anonymously at a light rate. Optional Google sign-in (OAuth 2.1) raises the limit — academic emails (.edu, .edu.au, .ac.*) get the highest tier. Every query is parameterised, row-capped and time-bounded. Data: COSMIC, DepMap/CCLE, PubMed.

Source

canelib_mcp.py — the tools + server. oauth.py — the OAuth 2.1 / tiering layer. Secrets live outside the repo (see .gitignore).

來源:README.md,提交 6aa4f36

工具

12

database_overview

Get the headline size of the CAN-IMMUNE database in one call. Returns total counts of mutations, genes, and unique mutant peptides, plus how many cell lines, tissues, and cancer types are covered, and the data sources (COSMIC, DepMap/CCLE, PubMed). Use this first to size the resource or to answer "how big is the database / how many X are there" questions. No parameters.

無需輸入參數。

search_genes

Find genes by symbol or full name, ranked by how mutated they are. Matches the gene symbol OR the full gene name (partial, case-insensitive) and returns each hit with its total mutation count, unique mutant-peptide count, sample count, and UniProt accession. Use it to locate a gene before calling `get_gene`, or to rank a set of genes by mutation burden. Results are ordered by mutation count (most mutated first).

輸入參數limitquery必填

get_gene

Get the full profile for one gene by exact symbol. Returns total mutations and how they split between cell lines vs tissues, unique mutant peptides, sample and transcript counts, UniProt accession / name / reviewed status, the top ~15 recurrent mutations in that gene, and a link to the gene page. Use this after `search_genes` (or when you already know the symbol) to answer detailed questions about a single gene. Returns an error field if the symbol is not found.

輸入參數symbol必填

search_cell_lines

Find cancer cell lines by name, ranked by mutation burden. Returns each matching cell line with its tissue of origin, cancer type, total mutations, unique mutant peptides, data sources (COSMIC / DepMap-CCLE / PubMed), and Cellosaurus ID. Covers only cell-line models (kept separate from primary tissue samples). Use before `get_cell_line` or `top_genes_in_cell_line`. Ordered by mutation count (most mutated first).

輸入參數limitquery必填

get_cell_line

Get the full profile for one cancer cell line by exact name. Returns tissue of origin, cancer type, total mutations, unique mutant peptides and genes, data sources, Cellosaurus and DepMap model IDs, gender, category / cell type, and the top ~15 most-mutated genes in that line, plus a link to its page. Use after `search_cell_lines`, or when the exact name is known, for a deep dive on one line. Returns an error field if the name is not found.

輸入參數name必填

top_genes_in_cell_line

Rank the most-mutated genes within one specific cell line. Answers "what is the top mutated gene in <cell line>" / "which genes are most altered in <cell line>". Returns genes with their mutation counts for that line, highest first, from precomputed per-cell-line stats (fast). For the whole profile of the line use `get_cell_line` instead.

輸入參數name必填limit

list_cancer_types

List cancer types (histology) ranked by mutation burden. Returns each cancer type / histology with its total mutations, unique samples, and unique genes, most-mutated first. Use to see which cancer types dominate the database or to pick one for further drill-down on the website.

輸入參數limit

list_tissues

List primary tissues ranked by mutation burden. Returns each tissue (COSMIC primary-tissue samples only, kept separate from cell lines) with its total mutations, unique samples, and unique genes, most-mutated first. Use for tissue-level questions; for cell lines use `search_cell_lines`.

輸入參數limit

get_gene_mutations

List individual mutations in a gene (a bounded sample of rows). Each row gives the CAN-IMMUNE mutation ID (CANLIB...), the amino-acid change, the CDS change, mutation type (missense / complex substitution), data source, the sample it came from, its primary site, and the mutant peptide. Ordered by mutation ID. Capped at `limit` rows (max 100) for speed - for the complete table of a heavily-mutated gene, link the user to the gene page on the website.

輸入參數gene必填limit

get_mhc_binding

Look up a CACHED MHC-I binding prediction for a peptide + HLA allele. Returns the NetMHCpan result (best binding core, %Rank_EL, binder level SB/WB/NB, and neoepitope %Rank_Neo when computed) for predictions already stored in the cache. This tool does NOT run new predictions - it only reads cached ones; if nothing is cached it returns cached=false with a note. To generate a new prediction, use the interactive button on the mutation page of the website.

輸入參數allele必填peptide必填

fetch

Fetch the full record for an id returned by `search` (ChatGPT connector interface). Takes an id like 'gene:TP53' or 'cell_line:MDA-MB-231' and returns {id, title, text, url, metadata} where `text` is the full JSON profile (same data as get_gene / get_cell_line). Always call `search` first to obtain a valid id.

輸入參數id必填

版本歷史

1
  1. v1.0.0最新Sep 16, 2026

評分

已滿足適用檢查項的 100%

該評分衡量此收錄條目的完整程度,依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價,也不表示伺服器安全或目前可用:可連通性是帶日期的檢查結果,評分也不包含安裝量或使用者評價。

  • 近 30 天內可連通必需25 分
  • 已宣告認證方式必需10 分
  • 已列出工具必需15 分
  • 描述必需15 分
  • 圖示10 分
  • 文件15 分
  • 授權條款10 分