Communitygithub.com

tamarind

Tamarind Bio 플랫폼에서 제공하는 오픈소스 분자 설계 및 구조 생물학 도구 모음에 REST API 또는 MCP 서버를 통해 액세스합니다. 로컬 GPU가 필요하지 않습니다.

tamarind란 무엇인가요?

이 스킬은 Tamarind Bio 클라우드 플랫폼과 상호작용하는 방법을 설명합니다. Tamarind Bio는 AlphaFold, Boltz, Chai, ESMFold, RFdiffusion, ProteinMPNN, DiffDock 등 수백 가지의 컴퓨터 생물학 도구를 관리형 GPU에서 실행합니다. 사용자는 로컬 하드웨어를 프로비저닝할 필요 없이 시퀀스나 구조를 제출하고 예측된 구조, 설계, 생물물리학적 점수를 받을 수 있습니다. 이 스킬은 인증, 도구 검색, 작업 제출, 결과 검색을 위한 두 가지 접근 방식(에이전트 호스트를 위한 MCP 서버와 보편적인 REST API)을 다룹니다. 또한 도구 선택, 작업 설정 검증, 파이프라인 체이닝, API 스키마가 자주 변경되므로 런타임에 라이브 소스를 가져와야 한다는 중요한 규칙에 대한 실용적인 지침을 제공합니다.

지원 대상~Claude Code~Codex CLI~Cursor✓Antigravity
npx skills add https://github.com/K-Dense-AI/scientific-agent-skills/tree/main/skills/tamarind

즐겨 사용하는 AI에게 물어보기

이 에이전트 스킬이 미리 로드된 새 채팅을 엽니다.

문서

Tamarind Bio

Tamarind Bio는 구조 예측, 단백질 및 항체 설계, 도킹, 결합 친화도, MSA 생성, 분자 동역학 등 컴퓨터 생물학 도구를 관리형 GPU에서 실행하는 클라우드 플랫폼입니다. 사용자는 시퀀스나 구조를 제출하고, 자체 하드웨어를 프로비저닝할 필요 없이 예측된 구조, 설계 및 생물물리학적 점수를 반환받습니다. 하나의 통일된 작업 API를 통해 수백 가지 도구(AlphaFold, Boltz-2, Chai-1, RFdiffusion, ProteinMPNN, BoltzGen, ESMFold2, DiffDock, Autodock Vina 등)를 제공합니다.

공식 문서: app.tamarind.bio/api-docs · 플랫폼 UI: app.tamarind.bio

정식 출처 — 이것들을 가져오고, 오래된 복사본에 의존하지 마세요

Tamarind는 라이브 기계 판독 가능 소스를 게시합니다. 하드코딩된 목록보다 런타임에 이를 가져오는 것을 선호하세요. 도구 이름, 스키마 및 엔드포인트는 자주 변경됩니다:

  • https://app.tamarind.bio/llms.txt — LLM 색인: 사양, API 문서 및 MCP 가이드 링크.
  • https://app.tamarind.bio/openapi.yaml — 8개의 핵심 작업 엔드포인트(submit-job/-batch, jobs, result, upload, files, delete-job/-file; 인증 ApiKeyAuth)에 대한 OpenAPI 3.0 사양입니다. 정확한 형태를 위해 가져오세요. 검색/관리 엔드포인트(/tools, /usage-statistics, 파이프라인 등)는 여기에 포함되어 있지 않습니다. 이러한 항목에는 MCP/REST 검색 도구를 사용하세요.
  • https://docs.tamarind.bio/llms.txt — 문서 색인; 모든 페이지에는 .md 형식이 있습니다 (예: docs.tamarind.bio/tamarind/batch.md, /tamarind/api.md, /tamarind/pipelines.md).
  • 라이브 도구 검색 — GET /tools (REST) 또는 MCP getAvailableTools + getJobSchema(jobType)은 존재하는 도구와 해당 매개변수에 대한 진실 공급원입니다.

이 스킬은 표면과 해당 출처에 명시되지 않은 명확하지 않은 동작을 가르칩니다 (참조 파일 참조). 모양에 대해 확신이 없으면 openapi.yaml을 가져오세요.

이 스킬을 사용해야 하는 경우

사용자가 다음을 원할 때 Tamarind를 사용하세요:

  • 단백질, 복합체 또는 단백질-리간드 시스템의 구조 예측 (AlphaFold, Boltz-2, Chai-1, ESMFold2, Chai/Boltz 공폴딩)
  • 단백질 또는 결합체 설계 (RFdiffusion, BoltzGen, BindCraft, ProteinMPNN/LigandMPNN 역폴딩)
  • 항체/나노바디 설계 또는 특성화 (시퀀스 생성, 인간화, 현실성, 면역원성)
  • 단백질에 소분자 도킹 (DiffDock, Autodock Vina) 또는 결합 친화도 예측
  • 다운스트림 폴딩을 위한 MSA 생성
  • 관리형 GPU에서 분자 동역학 또는 기타 생물물리학 워크플로 실행
  • 동일한 도구를 통해 많은 시퀀스나 설계 일괄 스크리닝
  • 파이프라인으로 도구 연결 (예: 설계 → 폴드 → 점수) 한 작업의 출력을 다음 작업의 입력으로 사용

이 스킬은 작업이 로컬 설치가 아닌 Tamarind의 관리형 클라우드에서 실행되어야 할 때 적합합니다. 순수 로컬 화학정보학 또는 일회성 시퀀스 I/O의 경우 대신 로컬 라이브러리(RDKit, BioPython)를 사용하세요.

액세스 및 인증

  1. app.tamarind.bio에 로그인하고 계정/API 설정에서 API 키를 생성합니다.
  2. 모든 REST 요청을 x-api-key 헤더로 인증합니다.
  3. 키를 절대 하드코딩하지 마세요. TAMARIND_API_KEY 환경 변수 또는 .env 파일에서 읽습니다 (python-dotenv 사용). 키를 소스 제어에 커밋하지 마세요.

가격: 모든 사용자는 10개의 무료 작업을 받습니다. 더 많은 사용량을 위해서는 [email protected]에 문의하여 구독을 구매하세요.

export TAMARIND_API_KEY="your_api_key"
# 사용 가능한 도구 나열
curl https://app.tamarind.bio/api/tools \
  -H "x-api-key: $TAMARIND_API_KEY"

기본 URL: https://app.tamarind.bio/api/

공식 Python SDK는 없습니다. tamarind라는 PyPI 패키지는 관련 없는 Neo4j 도구입니다. uv pip install tamarind를 실행하지 마세요. REST API에 대해 일반 requests 호출을 작성하거나(엔드포인트 모양은 openapi.yaml에 있음) 에이전트 호스트용 MCP 서버를 사용하세요.

Tamarind를 호출하는 두 가지 방법

MCP 서버 (AI 에이전트에 가장 적합)

Tamarind는 https://mcp.tamarind.bio/mcp에서 MCP 서버를 호스팅합니다 (X-API-Key 헤더를 통한 API 키 인증). 에이전트 호스트가 MCP를 지원하는 경우 선호하세요. 도구는 에이전트 친화적인 스키마로 REST API를 미러링합니다:

  • listModalities() / listTags() — 레이블 + 도구 개수가 포함된 라이브 필터 어휘(분자 유형/기능); 하드코딩 대신 유효한 modality/function 값을 배우려면 이들을 호출하세요
  • getAvailableTools(modality?, function?, search?, custom?) — 도구 검색 (category/tag는 여전히 지원되는 더 이상 사용되지 않는 별칭입니다)
  • getJobSchema(jobType) — 도구에 대한 정확한 매개변수 스키마 및 exampleJob 시작 페이로드 (제출하기 전에 검증)
  • validateJob(jobName, type, settings) — 제출하기 전 드라이런 검증
  • submitJob(jobName, type, settings) / submitBatch(batchName, type, settings[], jobNames[])
  • getJobs(jobName?, batch?, limit?, includeSequences?) — 작업 및 상태 나열/검사 (기본적으로 부피가 큰 작업별 입력 blob은 생략됨; 유지하려면 includeSequences=true 전달)
  • getJobLogs(jobName) — 디버깅을 위한 출력 로그 가져오기
  • listJobFiles(jobName) — 출력 파일 나열 (체이닝을 위해 s3Path 반환)
  • getResult(jobName, fileName?) — 결과 다운로드
  • uploadFile(filename) — 사전 서명된 업로드 URL; 또는 uploadFileContent(filename, content, encoding?)을 사용하여 호스트가 S3에 도달할 수 없을 때 MCP를 통해 파일 콘텐츠 전송 (샌드박싱된 에이전트)

범위 참고: MCP 쿼리 도구(getJobs, getResult, listJobFiles 등)는 인증된 계정으로 범위가 지정됩니다.

REST API (범용)

requests와 함께 일반 HTTP 사용 — 엔드포인트 모양은 openapi.yaml에 있습니다. 핵심 루프는 아래와 같습니다. references/workflows.md에 전체 레시피가 있습니다.

핵심 워크플로

항상 검색 → 스키마 → 검증 → 제출 → 폴링 → 결과 순서를 따르세요. 도구 이름이나 설정을 하드코딩하지 마세요. 카탈로그는 자주 변경됩니다.

import os, time, requests

BASE = "https://app.tamarind.bio/api"
HEADERS = {"x-api-key": os.environ["TAMARIND_API_KEY"]}

# 1. 도구 검색. REST /tools는 전체 목록을 반환합니다. 클라이언트 측에서 필터링합니다.
tools = requests.get(f"{BASE}/tools", headers=HEADERS).json()
alphafold = next(t for t in tools if t["name"] == "alphafold")

# 2. 선택한 도구에 대한 정확한 스키마를 가져옵니다.
#    REST: 각 /tools 항목에는 이미 인라인 `settings` 스키마(매개변수 목록)가 포함되어 있습니다. 이름이 작업 유형과 일치하는 항목을 찾으세요.
#    MCP:  getJobSchema(jobType)는 동일한 도구별 세부 정보를 반환합니다.

# 3. 작업을 제출합니다. `settings`는 도구별로 다릅니다. 스키마와 정확히 일치시킵니다.
payload = {
    "jobName": "my-alphafold-run",          # ^[a-zA-Z0-9_-]+$, <=100자, 고유해야 함
    "type": "alphafold",
    "settings": {
        "sequence": "MKTVRQERLKSIVRILERSKEPVSGAQLAEELSVSRQVIVQDIAYLRSLGYNIVATPRGYVLAGG",
        "numRecycles": 3,
    },
}
resp = requests.post(f"{BASE}/submit-job", headers=HEADERS, json=payload)
resp.raise_for_status()   # 200 성공; 400 잘못된 요청; 403 예산 초과; 401 인증되지 않음

# 4. 완료될 때까지 폴링합니다.
#    참고 응답 모양: GET /jobs?jobName=<name>은 직접 작업 ROW를 반환합니다 ("jobs" 래퍼 없음);
#    목록 쿼리(jobName 없음)는 {"jobs": [...]}를 반환합니다.
#    이름별 응답에서 ["jobs"][0]을 인덱싱하지 마세요.
while True:
    job = requests.get(f"{BASE}/jobs", headers=HEADERS,
                       params={"jobName": "my-alphafold-run"}).json()
    if job["JobStatus"] in ("Complete", "Stopped", "Deleted"):
        break
    time.sleep(30)

# 5. 결과 검색. POST /result는 사전 서명된 URL *문자열*을 반환합니다.
#    해당 URL을 GET하여 실제 결과 zip을 다운로드합니다 (2단계).
url = requests.post(f"{BASE}/result", headers=HEADERS,
                    json={"jobName": "my-alphafold-run"}).text.strip('"')
open("my-alphafold-run.zip", "wb").write(requests.get(url).content)

MCP 도구를 사용한 이 루프의 에이전트 버전 및 더 풍부한 예제는 references/workflows.md를 참조하세요.

도구 검색

카탈로그에는 수백 개의 도구가 있습니다. 항상 런타임에 열거하세요. 하드코딩된 목록에 절대 의존하지 마세요.

REST GET /tools는 전체 목록을 반환합니다 (서버 측에서 필터링하지 않음); 각 항목은 {name, displayName, github, paper, description, settings}이며, 여기서 settings는 해당 도구의 인라인 매개변수 스키마입니다. 클라이언트 측에서 필터링:

tools = requests.get(f"{BASE}/tools", headers=HEADERS).json()   # 목록
boltz = [t for t in tools if "boltz" in t["name"].lower()]

참고: 두 표면 모두 도구 이름당 하나의 행을 반환합니다. REST /tools와 MCP getAvailableTools는 모두 중복 제거됩니다 (MCP는 최신 도구 버전 유지). 따라서 이름 일치는 단일 행을 반환합니다.

MCP getAvailableTools(search=..., modality=..., function=...)는 서버 측에서 필터링하고 도구당 categories/tags를 추가합니다 (category/tag는 modality/function의 더 이상 사용되지 않는 별칭이며 여전히 지원됨). 어휘를 하드코딩하지 마세요. 변경됩니다. listModalities() / listTags()에서 라이브 값을 가져오거나 (각각 value, label, description 및 toolCount 반환) 모든 getAvailableTools 응답에서 반환된 availableCategories / availableTags 패싯 배열을 읽습니다. Modality는 분자 유형(단백질, 항체, 펩타이드, 소분자, 핵산 등)이고, function은 도구가 수행하는 작업(구조 예측, 결합체 설계, 단백질-리간드 도킹 등)입니다.

널리 사용되는 도구의 대표적인 세트 (/tools로 확인): alphafold, boltz (Boltz-2), chai (Chai-1), esmfold / esmfold2, rfdiffusion, proteinmpnn, ligandmpnn, boltzgen, bindcraft, diffdock. 전체 카테고리/태그 맵과 도구 메타데이터를 읽는 방법은 references/tool_catalog.md를 참조하세요.

올바른 도구 선택

카탈로그에는 작업당 많은 도구가 있습니다. 즐겨찾기를 하드코딩하지 마세요. function(및 modality)으로 필터링한 다음 각 후보의 description을 읽고 사용자의 실제 목표(가지고 있는 입력, 필요한 출력, 속도 또는 "MSA 없음"과 같은 제약 조건)와 일치시키세요. description 및 tags 필드는 공개 "용도" 신호입니다. 여기에 validateJob을 더하여 선택을 결정하세요. 작업별 빠른 방향:

  • 단일 단백질/복합체 폴드 (function=structure-prediction): AlphaFold3급 복제 — boltz/chai/openfold/protenix/intfold — 단백질 전용 시스템을 포함한 모든 것에 정확한 기본값입니다. 또한 핵산 + 소분자 복합체를 처리하므로 리간드/RNA/DNA가 시스템의 일부일 때 이들을 사용하세요 (그리고 boltz는 결합 친화도를 추가함). alphafold (AF2)는 단량체 + 다량체에 대해 여전히 견고한 선택입니다 (:로 체인 결합). esmfold는 단일 시퀀스(MSA 없음)이며 빠릅니다. 속도를 원하고 MSA가 없을 때 사용하세요. esmfold2는 더 새롭고 기본적으로 MSA를 조건으로 합니다 (model 설정은 더 빠른 단일 시퀀스 모드를 제공함). 항체(abodybuilder, immunebuilder), 고리형 펩타이드(highfold) 및 형태적 앙상블(afcluster, alphaflow)을 위한 특수 폴더가 있습니다. 필터링하고 설명을 읽으세요.
  • 결합체 설계 (function=binder-design): bindcraft (de novo 미니단백질 결합체) 및 boltzgen (단백질 및 소분자 표적에 대한 결합체, 나노바디/항체/펩타이드 포함)이 가장 적합한 de novo 결합체 도구입니다. rfdiffusion도 결합체 설계를 수행하며 모티프 스캐폴딩 / 기존 백본 다양화를 위한 선택입니다. 항체별 생성기는 function=antibody-design 아래에 있습니다.
  • 알려진 백본에 대한 시퀀스 설계 (function=inverse-folding): proteinmpnn (일반), ligandmpnn (리간드 인식), 내열성/가용성/항체 MPNN 변종. 역폴딩은 구조를 받아 시퀀스를 출력합니다. 이를 다시 폴드하여 확인합니다(체이닝 참조).
  • 소분자 도킹 (function=protein-ligand-docking): boltz/chai를 선호하세요. 이들은 복합체에 리간드를 공폴딩하고 고정된 수용체에 도킹하는 대신 결합된 구조를 예측합니다. 알려진 포켓에 대한 빠르고 대규모 스크리닝이 필요할 때 autodock-vina를 사용하세요.
  • 결합 친화도 예측 (function=binding-affinity) 또는 MSA 생성 (msa 검색) — 필터링하고 읽으세요.

사용자가 특정 도구를 지명하면 해당 도구를 평가 하고 해당 tag 그룹의 대안을 검증하세요. 더 빠르거나 더 적절한 형제가 종종 존재합니다. 확실하지 않으면 getJobSchema/validateJob을 사용하여 후보가 실제로 가지고 있는 입력을 수락하는지 확인한 후 제출하세요.

작업 설정, 스키마 및 검증

각 도구에는 고유한 settings 스키마가 있습니다. 제출하기 전에 가져오세요:

  • REST /tools 항목: 각 settings 매개변수는 트리밍된 딕셔너리입니다. name과 required만 항상 존재합니다. type, default, description, options는 관련이 있을 때만 나타납니다 (약 60%에 type이 있음). 따라서 param.get("type")를 사용하고 param["type"]을 사용하지 마세요. 고급 게이팅 키(exclude, conditionals)는 REST 응답에 전혀 없습니다.
  • MCP getJobSchema(jobType): 이러한 게이팅 키에 대해 추론해야 할 때 MCP를 사용하세요. (restrictOrgs는 두 표면에서 제거됩니다. 사용할 수 없는 조직 게이팅 매개변수는 단순히 생략됩니다. references/api_reference.md 참조.)

제출하기 전에 항상 validateJob (MCP)하세요 — 신뢰할 수 있는 가드입니다. 제출하지 않고 /submit-job과 동일한 검증을 실행하고 첫 번째 누락/잘못된 필드를 표시합니다. 스키마 키에서 제거할 필드를 수동으로 유도하려고 하지 마세요 (어차피 REST에서는 볼 수 없음) — validateJob이 알려줍니다. (응답에는 예를 들어 "static-fallback"과 같은 source 필드가 포함될 수 있습니다. 이는 검증된 스키마 소스에 대한 내부 참고 사항입니다. valid: true/false는 당신이 행동하는 신호입니다.)

validateJob은 기본값이 채워진 설정의 normalized 보기를 에코합니다. 검증한 동일한 깨끗한 settings를 제출하세요. normalized를 정보용으로 취급하세요 (사용자가 설정하지 않은 기본값과 일부 도구의 플랫폼 관리 필드를 전달할 수 있음). 따라서 정규화된 blob이 아닌 자체 설정에서 제출을 구성하세요.

시퀀스: 아미노산 문자열; 다량체의 사슬을 구분하려면 :를 사용하세요. 예: "MVLSPADKTNVKAAWGKVGAHAGEYGAEALERMFLSFPTTKTYFPHF...:MVLSPADKTNVKAAWGKVGAHAGEYGAEALERMFLSFPTTKTYFPHF...". getJobSchema를 확인하여 각 도구가 다중 사슬 입력을 수락하는 방법을 확인하세요.

결과 체이닝 및 파이프라인

한 작업의 출력을 다음 작업의 입력으로 사용할 때 항상 input / inputs 키를 통해 s3Path를 전달하고 직접 내용을 복사하지 마세요 — 큰 파일의 경우 비효율적이고 불안정합니다. listJobFiles는 각 출력 파일에 대해 s3Path를 반환합니다. 다음과 같이 설정에서 input(들)을 전달하세요:

# 구조 예측 작업의 출력 파일 나열
files = client.call_tool("listJobFiles", {"jobName": "my-fold-job"})
pdb_path = [f for f in files if f["fileName"].endswith(".pdb")][0]["s3Path"]

# 역폴딩 작업 제출, 입력으로 s3Path 사용
client.call_tool("submitJob", {
    "jobName": "my-design",
    "type": "proteinmpnn",
    "settings": {
        "input": pdb_path,    # <-- 내용이 아닌 S3 경로
        "numSeq": 1,
    }
})

작업 유형에 따라 input 또는 inputs 키가 있을 수 있습니다. 스키마를 확인하세요. s3Path 접두사 s3://tamarind-prod-... 또는 s3://mason-...일 수 있습니다. 상관없습니다. URL은 플랫폼 내에서 확인되며 전달할 수 있습니다 (단, 외부 다운로드에는 작동하지 않으므로 시도하지 마세요).

파이프라인은 references/workflows.md에서 더 자세히 다룹니다.

윤리 메모

Tamarind는 생물의학 연구 및 약물 발견을 가속화하기 위한 합성 생물학 도구를 제공합니다. 윤리적인 사용 (독성 물질, 병원체 또는 무기화에 대한 작업 없음)을 장려하세요. 질문이 있는 경우 [email protected]로 연락하세요.

Individual skills in this repo

This repo contains 20 individual skills — each has its own dedicated page.

adaptyv

How to use the Adaptyv Bio Foundry API and Python SDK for protein experiment design, submission, and results retrieval. Use this skill whenever the user mentions Adaptyv, Foundry API, protein binding assays, protein screening experiments, BLI/SPR assays, thermostability assays, or wants to submit protein sequences for experimental characterization. Also trigger when code imports `adaptyv`, `adaptyv_sdk`, or `FoundryClient`, or references `foundry-api-public.adaptyvbio.com`.

aeon

This skill should be used for time series machine learning tasks including classification, regression, clustering, forecasting, anomaly detection, segmentation, and similarity search. Use when working with temporal data, sequential patterns, or time-indexed observations requiring specialized algorithms beyond standard ML approaches. Particularly suited for univariate and multivariate time series analysis with scikit-learn compatible APIs.

alphagenome

Look up precomputed AlphaGenome Atlas effects for any GRCh38 single-nucleotide variant (AVI score with Phred and 18 SHAP feature attributions, plus raw and quantile scores for RNA-seq, DNase, ATAC, ChIP-TF, ChIP-histone, CAGE, PRO-cap, splicing, polyadenylation and contact-map tracks), score variants or scan windows on demand with the AlphaGenome model for human and mouse (variant scoring, in silico mutagenesis, REF-versus-ALT track prediction), and build Atlas website deep links. Use when the user mentions AlphaGenome, AlphaGenome Atlas, AVI or AlphaGenome Variant Impact, DeepMind variant effect prediction, or wants to prioritise or mechanistically interpret non-coding, regulatory, splicing, enhancer, promoter, or chromatin-accessibility effects of SNVs from a VCF, credible set, or region. Research use only; not a clinical tool.

analytical-method-validation

Plan, execute, and document validation, verification, and transfer of analytical procedures under the governing framework - ICH Q2(R2) and Q14, USP <1220>/<1225>/<1226>, ICH M10 bioanalytical, CLSI EP, or ISO/IEC 17025. Use for HPLC, LC-MS/MS, GC, CE, ICP-MS, dissolution, qNMR, qPCR, NIR, and ligand binding or cell-based assays whenever the question is whether a procedure is fit for its intended purpose. Triggers include

anndata

Data structure for annotated matrices in single-cell analysis. Use when working with .h5ad files or integrating with the scverse ecosystem. This is the data format skill—for analysis workflows use scanpy; for probabilistic models use scvi-tools; for population-scale queries use cellxgene-census.

arbor

Autonomously improve a real artifact (code, training recipe, agent harness, data pipeline, prompt) against an objective and an evaluator, using Hypothesis Tree Refinement (HTR) from the Arbor paper. Use this whenever someone wants to iteratively optimize something over many experiments without overfitting — e.g.

arboreto

Infer gene regulatory networks (GRNs) from gene expression data using scalable algorithms (GRNBoost2, GENIE3). Use when analyzing transcriptomics data (bulk RNA-seq, single-cell RNA-seq) to identify transcription factor-target gene relationships and regulatory interactions. Supports distributed computation for large-scale datasets.

astropy

Core Python library for astronomy and astrophysics workflows that need Astropy APIs, including units/quantities, coordinates, FITS I/O, tables, time systems, WCS, and cosmology. Use when implementing or debugging astronomical data analysis code with Astropy.

autoskill

Observe the user

benchling-integration

Benchling Python SDK and REST API integration for registry entities, inventory, ELN entries, workflows, Benchling Apps, and Data Warehouse queries. Use when automating lab data with benchling-sdk or the v2 API.

bgpt-paper-search

Search scientific papers and retrieve structured experimental data extracted from full-text studies via the BGPT MCP server. Returns 25+ fields per paper including methods, results, sample sizes, quality scores, and conclusions. Use for literature reviews, evidence synthesis, and finding experimental details not available in abstracts alone.

bids

>

biopython

Comprehensive molecular biology toolkit. Use for sequence manipulation, file parsing (FASTA/GenBank/PDB), phylogenetics, and programmatic NCBI/PubMed access (Bio.Entrez). Best for batch processing, custom bioinformatics pipelines, BLAST automation. For quick lookups use gget; for multi-service integration use bioservices.

bioservices

Unified Python interface to 40+ bioinformatics services. Use when querying multiple databases (UniProt, KEGG, ChEMBL, Reactome) in a single workflow with consistent API. Best for cross-database analysis, ID mapping across services. For quick single-database lookups use gget; for sequence/file manipulation use biopython.

bulk-rnaseq

End-to-end bulk RNA-seq orchestrator — takes raw FASTQ reads through QC and trimming (FastQC, fastp/Trim Galore), alignment and quantification (STAR, Salmon, featureCounts), assembles a gene-level counts matrix, then hands off to differential expression (pydeseq2), pathway/GSEA enrichment (pathway-enrichment), and publication figures (scientific-visualization). Use whenever the user has bulk RNA-seq reads or quant output and wants a complete, reproducible differential-expression workflow — e.g.

cellxgene-census

Query the CZ CELLxGENE Census programmatically for versioned public single-cell and spatial transcriptomics data. Use when you need population-scale cell metadata, gene expression slices, Census summary counts, source H5AD URIs/downloads, embeddings, spatial Census data, or reference atlas comparisons across organisms, tissues, diseases, assays, and cell types. For analyzing your own local single-cell data use scanpy, anndata, or scvi-tools.

cirq

Google quantum computing framework. Use when targeting Google Quantum AI hardware, designing noise-aware circuits, or running quantum characterization experiments. Best for Google hardware, noise modeling, and low-level circuit design. For IBM hardware use qiskit; for quantum ML with autodiff use pennylane; for physics simulations use qutip.

citation-management

Comprehensive citation management for academic research. Search OpenAlex, PubMed, and Google Scholar for papers, extract accurate metadata, validate citations, and generate properly formatted BibTeX entries. This skill should be used when you need to find papers, verify citation information, convert DOIs to BibTeX, or ensure reference accuracy in scientific writing.

clinical-decision-support

Prepare and validate research-only clinical decision-support evaluation, evidence-profile, cohort, survival, biomarker/model, privacy, and governance artifacts. Use for aggregate or synthetic research documentation and traceability—not patient care or live clinical operation.

clinical-reports

Create safety-bounded draft structures and run local deterministic checks for clinical case, diagnostic, trial, safety, and aggregate research reports. Use only with synthetic, de-identified, or aggregate inputs and verified source-fact manifests; every output requires qualified review.

관련 스킬