К основному содержанию
скилл CX Codex

tooluniverse-admet-prediction — скилл для Claude Code

Скилл tooluniverse-admet-prediction для Claude Code профилирует кандидатов в лекарства по ADMET-свойствам. До лабораторных тестов помогает отсеять заведомо проблемные соединения.

Apache-2.0 245 установок ★ 1 798 обновлён 2026-10-24

Установить в Claude Code · Через Skills CLI (Vercel) Все способы и платформы ↓
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-admet-prediction -a claude-code
Открыть на GitHub

Что делает tooluniverse-admet-prediction

Скилл объединяет предсказания ADMET-AI, проверку drug-likeness по SwissADME, экспериментальную токсичность из PubChemTox, клинические данные ChEMBL, правило пяти Липински и сведения о CYP-взаимодействиях. По названию препарата или SMILES-строке он проходит пять фаз: идентификация соединения через PubChem, физико-химические свойства и drug-likeness, ADME-предсказания, оценка токсичности, итоговая карта pass/warn/fail с клиническим контекстом.

Ключевое правило скилла — сначала вычислить, потом описывать. Если нужен статистический анализ или обсчёт, агент пишет Python-код и запускает его через Bash, а не рассуждает о том, что получилось бы. Каждое свойство оценивается отдельно: хорошая всасываемость не компенсирует гепатотоксичность. Приоритет у экспериментальных данных (T2) над вычислительными (T3), а любой FAIL в токсичности (hERG, AMES, DILI) считается ограничивающим для программы, пока wet-lab не опровергнет.

Для работы ADMETAI-инструментов нужен пакет tooluniverse[ml]: uv pip install 'tooluniverse[ml]'. Без него инструменты видны в tu list, но падают при вызове с ошибкой ADMETModel requires 'admet-ai' package. Первый вызов ADMETAI грузит PyTorch и печатает предупреждения о GPU и TypedStorage — это шум библиотек при CPU-инференсе, не сбой. Ошибкой считается только поле error в ответе или отсутствие предсказаний.

Какие задачи решает

  • проверить, нарушает ли соединение правило пяти Липински
  • оценить проницаемость гематоэнцефалического барьера
  • предсказать гепатотоксичность и hERG-риски до лаборатории
  • собрать ADME/PK-профиль по SMILES или названию препарата
  • отсеять проблемные молекулы при скрининге библиотеки

Примеры запросов агенту

  • Какие ADMET-свойства у ибупрофена?
  • Пройдёт ли это соединение через гематоэнцефалический барьер: CC(C)Cc1ccc(cc1)C(C)C(=O)O
  • Проверь эту молекулу по правилу пяти Липински

Как установить tooluniverse-admet-prediction

1Через Skills CLI (Vercel)

npx skills add mims-harvard/ToolUniverse --skill tooluniverse-admet-prediction -a claude-code

2Вручную: положить файлы

~/.claude/skills/tooluniverse-admet-prediction/SKILL.md

Папка скилла целиком; для одного проекта — .claude/skills/tooluniverse-admet-prediction/

По шагам, с командами для каждой платформы: Как установить скилл в Claude Code: пять способов

Вопросы о tooluniverse-admet-prediction

Как установить tooluniverse-admet-prediction в Claude Code?

Выполни npx skills add mims-harvard/ToolUniverse --skill tooluniverse-admet-prediction -a claude-code. Скилл появится в ~/.claude/skills/tooluniverse-admet-prediction/SKILL.md.

Что нужно для работы скилла?

ADMETAI-инструменты запускают локальную модель и требуют пакет tooluniverse[ml]: uv pip install 'tooluniverse[ml]'. Без него вызовы падают с ошибкой ADMETModel requires 'admet-ai' package. Проверить установленные группы можно через tooluniverse-doctor.

Чем скилл отличается от простого запроса к агенту?

Он заставляет агента сначала резолвить идентичность соединения через PubChem, а не угадывать SMILES или CID. Все ADMETAI-инструменты принимают smiles: list[str], SwissADME — одиночную строку, а PubChemTox — CID или название. Каждый анализ заканчивается картой pass/warn/fail с грейдированием доказательств.

Исходный текст

SKILL.md · Apache-2.0
---
name: tooluniverse-admet-prediction
description: Comprehensive ADMET (Absorption, Distribution, Metabolism, Excretion, Toxicity) profiling for drug candidates. Integrates ADMET-AI predictions, SwissADME drug-likeness, PubChemTox experimental toxicity, ChEMBL clinical data, Lipinski rule-of-five, and CYP interaction data. Use for drug-likeness assessment, BBB penetration, bioavailability, hepatotoxicity prediction, ADME/PK profiling, or screening compound libraries before lab testing.
disable-model-invocation: true
---

# ADMET Prediction & Drug Candidate Profiling

**ADMET reasoning**: a drug fails if it can't be absorbed, distributes to wrong tissues, isn't metabolized safely, or isn't excreted. Evaluate each property independently — good absorption doesn't compensate for liver toxicity. The ADME properties determine whether a compound reaches its target at therapeutic concentrations; toxicity determines whether it's safe to do so. Prioritize experimental data (T2) over computational predictions (T3) — ADMETAI predictions are screening tools, not definitive verdicts. When a FAIL is flagged in any toxicity category (hERG, AMES, DILI), treat it as program-limiting until wet-lab data refutes it.

**LOOK UP DON'T GUESS**: never assume SMILES, CID, or experimental LD50 values — always call PubChem to resolve compound identity before any ADMETAI or PubChemTox call.

Comprehensive pharmacokinetic and toxicity profiling integrating AI-based ADMET predictions, rule-based drug-likeness filters, and experimental benchmarks from curated databases.

## When to Use This Skill

**Triggers**:
- "What are the ADMET properties of [compound]?"
- "Is [drug] likely to cross the blood-brain barrier?"
- "Predict the toxicity of this SMILES: ..."
- "Does [compound] violate Lipinski's rule of five?"
- "Assess the drug-likeness of [molecule]"
- "What are the CYP interactions for [drug]?"
- "Pharmacokinetic profile of [compound]"
- "Is [compound] orally bioavailable?"
- "What is the LD50 / hERG liability of [molecule]?"

**Input**: Drug name (e.g., "ibuprofen") OR SMILES string (e.g., "CC(C)Cc1ccc(cc1)C(C)C(=O)O")

## Before You Run

ADMETAI tools run a local model, so they need the `ml` extra:

```bash
uv pip install 'tooluniverse[ml]'
```

Without it the tools still appear in `tu list` (the config loads) but fail at
call time with `ADMETModel requires 'admet-ai' package`. Run
`tooluniverse-doctor` to confirm which optional groups are installed.

**Expected console noise — not errors.** The first ADMETAI call loads PyTorch
and prints warnings such as missing-GPU / `Trainer` messages from
PyTorch Lightning, and `TypedStorage is deprecated` from PyTorch. These are
emitted by the underlying libraries during normal CPU inference. Predictions
are unaffected — do not report them to the user as failures and do not retry
the call because of them. Only treat output as a failure if the tool returns an
`error` field or no predictions.

---

## COMPUTE, DON'T DESCRIBE
When analysis requires computation (statistics, data processing, scoring, enrichment), write and run Python code via Bash. Don't describe what you would do — execute it and report actual results. Use ToolUniverse tools to retrieve data, then Python (pandas, scipy, statsmodels, matplotlib) to analyze it.

## KEY PRINCIPLES

1. **Resolve identity first** - Always convert drug name to SMILES before calling ADMETAI tools
2. **ADMETAI tools require `tooluniverse[ml]`** - If import fails, skip to SwissADME/PubChemTox fallbacks
3. **All ADMETAI tools take `smiles: list[str]`** - Always wrap in a list, even for one compound
4. **SwissADME takes `smiles: str`** - Single string, NOT a list (SOAP-style with `operation` param)
5. **PubChemTox tools accept `cid` or `compound_name`** - Use CID when available for reliability
6. **Evidence grading mandatory** - Predictions (T3), experimental data (T2), regulatory (T1)
7. **Scorecard output** - Every analysis must end with a pass/warn/fail scorecard
8. **Explain significance** - State WHY each property matters for drug development

---

## Evidence Grading

| Tier | Label | Source |
|------|-------|--------|
| **T1** | Regulatory/Clinical | FDA labels, ChEMBL max clinical phase |
| **T2** | Experimental | PubChemTox LD50/LC50, in vitro AMES, animal studies |
| **T3** | Computational | ADMETAI predictions, SwissADME calculations |
| **T4** | Annotation | Database cross-references, text-mined |

## Workflow: 5-Phase ADMET Profiling

```
User Query (drug name or SMILES)
|
+-- PHASE 1: Compound Identity Resolution
|   PubChem name->CID->SMILES, or validate input SMILES
|
+-- PHASE 2: Physicochemical & Drug-Likeness
|   ADMETAI physicochemical + SwissADME druglikeness -> Lipinski/Veber
|
+-- PHASE 3: ADME Predictions
|   BBB, bioavailability, CYP interactions, clearance, solubility
|
+-- PHASE 4: Toxicity Assessment
|   ADMETAI tox + PubChemTox experimental + nuclear receptor + stress
|
+-- PHASE 5: Scorecard & Clinical Context
|   ChEMBL max phase, aggregate pass/warn/fail, final recommendation
```

---

### PHASE 1: Compound Identity Resolution

**Goal**: Obtain SMILES, PubChem CID, and basic identifiers for the query compound.

**Steps**:

1. **If input is a drug name**:
   - Call `PubChem_get_CID_by_compound_name(name=<drug_name>)` to get CID
   - Call `PubChem_get_compound_properties_by_CID(cid=<CID>)` to get SMILES and MW
   - Extract `ConnectivitySMILES` from the response (NOT `CanonicalSMILES`)

2. **If input is a SMILES string**:
   - Call `PubChem_get_CID_by_SMILES(smiles=<SMILES>)` to get CID
   - Call `PubChem_get_compound_properties_by_CID(cid=<CID>)` for compound name and MW
   - Use the input SMILES for all subsequent ADMETAI calls

3. **Record**:
   - Compound name, CID, SMILES, molecular formula, molecular weight, IUPAC name
   - If CID lookup fails, proceed with SMILES only (ADMETAI does not need CID)

**Why this matters**: ADMETAI tools require SMILES input. PubChemTox tools work best with CID. Resolving both ensures all downstream tools can be called. PubChem is the authoritative source for SMILES canonicalization.

**Fallback**: If PubChem has no entry, the user must provide SMILES directly. Cannot proceed without SMILES.

---

### PHASE 2: Physicochemical Properties & Drug-Likeness

**Goal**: Evaluate whether the compound has drug-like physicochemical properties.

**Steps**:

1. **ADMETAI physicochemical** (primary):
   ```
   ADMETAI_predict_physicochemical_properties(smiles=["<SMILES>"])
   ```
   Returns: MW, logP, TPSA, HBD, HBA, rotatable bonds

2. **SwissADME drug-likeness** (complementary):
   ```
   SwissADME_check_druglikeness(operation="check_druglikeness", smiles="<SMILES>")
   SwissADME_calculate_adme(operation="calculate_adme", smiles="<SMILES>")
   ```
   Returns: Lipinski, Veber, Ghose, Egan, Muegge rule compliance; PAINS alerts; Brenk alerts

3. **ADMETAI solubility**:
   ```
   ADMETAI_predict_solubility_lipophilicity_hydration(smiles=["<SMILES>"])
   ```
   Returns: Aqueous solubility (LogS), lipophilicity, hydration free energy

**Interpret & Score**:

| Property | Ideal Range | Why It Matters |
|----------|-------------|----------------|
| MW | < 500 Da | Larger molecules have poor membrane permeability (Lipinski) |
| LogP | -0.4 to 5.6 | Too hydrophobic = poor solubility; too hydrophilic = poor permeability |
| HBD | <= 5 | Excess donors reduce membrane crossing (Lipinski) |
| HBA | <= 10 | Excess acceptors reduce membrane crossing (Lipinski) |
| TPSA | < 140 A^2 | High PSA correlates with poor oral absorption |
| Rotatable bonds | <= 10 | Molecular flexibility affects bioavailability (Veber) |
| LogS | > -6 | Below -6 = practically insoluble, formulation challenge |
| PAINS alerts | 0 | Pan-assay interference compounds give false positives in screens |

**Verdict**: PASS if Lipinski <= 1 violation and no PAINS alerts; WARN if 2 violations; FAIL if 3+ violations or PAINS+.

**Fallback**: If ADMETAI import fails (missing `tooluniverse[ml]`), rely on SwissADME alone. SwissADME provides all Lipinski descriptors independently.

**Local TDC oracle scoring (no ML extra, no PubChem round-trip needed)**: `TDC_predict_oracle_score(smiles=..., oracle=...)` computes drug-likeness/property scores locally via the PyTDC package straight from a SMILES string — useful when `tooluniverse[ml]` isn't installed or as a fast independent cross-check. Six oracles: `QED` (drug-likeness 0-1), `SA` (synthetic accessibility 1=easy to 10=hard), `LogP`, and three target-binding-likelihood oracles (`GSK3B`, `JNK3`, `DRD2`) useful for early candidate triage against those specific targets. **Verified live and cross-checked**: `TDC_predict_oracle_score(smiles="CC(=O)Oc1ccccc1C(=O)O", oracle="SA")` and the independent `Chem_sa_score(smiles=...)` tool both returned **1.580** for aspirin — same underlying RDKit SA_Score algorithm, two separate wrapper tools; either is fine to cite, don't report both as independent corroborating evidence since they're the same computation.

`TDC_load_dataset(problem=..., name=...)` / `TDC_list_datasets(problem=...)` give access to TDC's ADME/Tox/HTS benchmark datasets — relevant if validating a custom prediction model against a public benchmark, not part of a single-compound ADMET profile. **`TDC_list_datasets` works** (verified live: `problem="ADME"` → 27 real dataset names). **`TDC_load_dataset` is currently broken** — verified live and reproduced on two different datasets (`Caco2_Wang`, `hERG`): it downloads successfully but the loader then receives an HTML error page instead of data and raises `Please report this error to [email protected]`. Don't build a workflow around it; use `TDC_list_datasets` for discovery and note the load failure if a user specifically needs a dataset pulled.

---

### PHASE 3: ADME Predictions

**Goal**: Predict absorption, distribution, metabolism, and excretion behavior.

**Steps**:

1. **Blood-brain barrier penetration**:
   ```
   ADMETAI_predict_BBB_penetrance(smiles=["<SMILES>"])
   ```
   - BBB+ = compound can cross; BBB- = cannot
   - Critical for CNS drugs (must cross) and peripherally-acting drugs (should NOT cross to avoid CNS side effects)

2. **Oral bioavailability**:
   ```
   ADMETAI_predict_bioavailability(smiles=["<SMILES>"])
   ```
   - F20% = at least 20% oral bioavailability; F30% = at least 30%
   - Low bioavailability means the drug is extensively metabolized or poorly absorbed
   - F < 20% generally requires non-oral routes (IV, inhaled, topical)

3. **CYP450 interactions**:
   ```
   ADMETAI_predict_CYP_interactions(smiles=["<SMILES>"])
   ```
   - Reports substrate/inhibitor status for CYP1A2, 2C9, 2C19, 2D6, 3A4
   - **Why CYP matters**: ~75% of drugs are metabolized by CYP enzymes. Inhibiting CYP3A4 (which metabolizes ~50% of drugs) causes dangerous drug-drug interactions (DDIs). CYP2D6 polymorphisms affect ~25% of drugs -- poor metabolizers accumulate toxic levels
   - Substrate of CYP2D6 = pharmacogenomic risk (poor/ultra-rapid metabolizers)
   - Inhibitor of CYP3A4 = high DDI risk (co-administered drugs accumulate)

4. **Clearance and distribution**:
   ```
   ADMETAI_predict_clearance_distribution(smiles=["<SMILES>"])
   ```
   - VDss (volume of distribution): low (<0.7 L/kg) = confined to plasma; high (>1 L/kg) = distributed to tissues
   - Clearance: high clearance = short half-life, frequent dosing needed
   - Plasma protein binding (PPB): >95% bound = narrow therapeutic window, DDI risk from displacement

5. **SwissADME pharmacokinetics** (cross-validation):
   - GI absorption (high/low), P-gp substrate status, skin permeation (logKp)

**Key flags**: BBB+ for non-CNS drug (WARN: CNS side effects); BBB- for CNS drug (FAIL: won't reach target); F < 20% (WARN: poor oral bioavailability); CYP3A4 inhibitor (WARN: high DDI); CYP2D6 substrate (WARN: pharmacogenomic variability); PPB > 99% (WARN: narrow window); high clearance + low bioavailability (FAIL).

**Fallback**: If ADMETAI unavailable, SwissADME provides GI absorption, BBB permeation (yes/no), P-gp substrate, and CYP inhibition predictions.

---

### PHASE 4: Toxicity Assessment

**Goal**: Evaluate safety liabilities from both predicted and experimental sources.

**Steps**:

1. **ADMETAI toxicity predictions** [T3]:
   ```
   ADMETAI_predict_toxicity(smiles=["<SMILES>"])
   ```
   Key endpoints:
   - **AMES**: Mutagenicity (bacterial reverse mutation test). Positive = potential carcinogen; regulatory agencies require AMES testing for all new drugs
   - **DILI**: Drug-induced liver injury risk. Leading cause of drug withdrawal (e.g., troglitazone). Positive = hepatotoxicity concern requiring liver function monitoring
   - **hERG**: hERG potassium channel inhibition. Causes QT prolongation and fatal cardiac arrhythmia. hERG+ = cardiotoxicity liability; multiple drugs withdrawn for this (e.g., terfenadine, cisapride)
   - **ClinTox**: Clinical trial toxicity / FDA withdrawal risk. Trained on drugs that failed trials or were withdrawn for toxicity
   - **LD50_Zhu**: Predicted lethal dose (mg/kg, rat oral). Lower = more acutely toxic
   - **Skin_Reaction**: Dermal sensitization potential. Important for topical drugs
   - **Carcinogens_Lagunin**: Carcinogenicity prediction

2. **Nuclear receptor activity** [T3]:
   ```
   ADMETAI_predict_nuclear_receptor_activity(smiles=["<SMILES>"])
   ```
   - AR (androgen receptor), ER (estrogen receptor), AhR, PPAR-gamma activity
   - Positive = potential endocrine disruption; critical for chronic-use drugs and environmental chemicals

3. **Stress response pathways** [T3]:
   ```
   ADMETAI_predict_stress_response(smiles=["<SMILES>"])
   ```
   - p53 activation = DNA damage response (genotoxicity signal)
   - MMP disruption = mitochondrial toxicity
   - ATAD5 = DNA repair stress
   - HSE = heat shock / protein misfolding stress

4. **PubChemTox experimental data** [T2] (call all in parallel):
   ```
   PubChemTox_get_toxicity_values(cid=<CID>)
   PubChemTox_get_ghs_classification(cid=<CID>)
   PubChemTox_get_acute_effects(cid=<CID>)
   PubChemTox_get_carcinogen_classification(cid=<CID>)
   PubChemTox_get_target_organs(cid=<CID>)
   PubChemTox_get_toxicity_summary(cid=<CID>)
   ```
   - Real animal study data (LD50, LC50, NOAEL) anchors computational predictions
   - GHS classification provides internationally harmonized hazard categories
   - Carcinogen classification from IARC (Group 1/2A/2B), NTP, EPA

**Key flags**: AMES positive (FAIL: mutagenic); DILI positive (WARN: hepatotox); hERG positive (FAIL: cardiac, often program-killing); ClinTox positive (WARN); LD50 < 50 mg/kg (FAIL: GHS 1-2); LD50 50-300 mg/kg (WARN: GHS 3); NR-ER/AR active (WARN: endocrine disruption); p53 active (WARN: genotoxicity); IARC Group 1/2A (FAIL: known/probable carcinogen).

**Fallback**: If ADMETAI unavailable, PubChemTox provides experimental toxicity data for known compounds. For novel compounds without PubChem entries, flag as "no experimental toxicity data available -- computational predictions only."

---

### PHASE 5: Scorecard Assembly & Clinical Context

**Goal**: Aggregate all findings into a structured ADMET scorecard with pass/warn/fail verdicts.

**Steps**:

1. **ChEMBL clinical status** [T1] (if drug has ChEMBL ID):
   ```
   ChEMBL_get_molecule(chembl_id="<CHEMBL_ID>")
   ```
   - Max phase: 4 = approved, 3 = Phase III, 2 = Phase II, 1 = Phase I, 0 = preclinical
   - Ro5 violations from ChEMBL (independent validation of Lipinski)
   - First approval year, indication class, black box warning flag

2. **Build the ADMET Scorecard**: produce a table with 13 categories (Physicochemical, Solubility, Absorption, Distribution, Metabolism, Excretion, Tox: Mutagenicity/Hepatotoxicity/Cardiotoxicity/Carcinogenicity/Acute, Endocrine, Clinical Tox), each with PASS/WARN/FAIL verdict and key finding. Include compound identity header and overall verdict. Tag each finding with evidence tier [T1-T3].

3. **Interpretation narrative**: After the scorecard, provide a 3-5 sentence summary:
   - Highlight the most critical findings (any FAILs or WARNs)
   - State whether the compound is suitable for oral administration
   - Note any DDI risks from CYP interactions
   - Flag pharmacogenomic concerns (CYP2D6 substrate)
   - Recommend next steps (e.g., "hERG patch clamp assay recommended to confirm computational prediction")

---

## Completeness Checklist (MANDATORY before reporting)

Before delivering the final scorecard, verify:

- [ ] Compound identity resolved (name, CID, SMILES all present or explicitly noted as unavailable)
- [ ] Physicochemical properties reported with Lipinski verdict
- [ ] At least one source for each ADME property (ADMETAI or SwissADME)
- [ ] All 7 ADMETAI toxicity endpoints reported (or marked N/A with reason)
- [ ] PubChemTox experimental data checked (even if "no data found")
- [ ] Nuclear receptor and stress response checked (or marked N/A)
- [ ] Evidence tier tagged for every finding
- [ ] Scorecard table complete with verdicts for all 13 categories
- [ ] Overall verdict stated
- [ ] Interpretation narrative provided with actionable next steps

© mims-harvard · Apache-2.0 · mims-harvard/ToolUniverse

Похожие скиллы

Все скиллы по теме «Наука и исследования» →
pubchem-database google-deepmind Скилл pubchem-database для работы с базой PubChem: ищет соединения по имени, CID или SMILES и вытаскивает свойства, безопасность и биоактивность. CX Codex +1 2 400установок deepchem K-Dense-AI Настраивает пайплайны хемоинформатики и предсказания свойств молекул в DeepChem 2.8.0. Помогает подготовить SMILES и обучить модели ADMET. CX Codex +1 2 100установок torchdrug K-Dense-AI Помогает настроить пайплайны TorchDrug 0.2.1 для графов молекул и белков. Пригодится при сборке моделей прогнозирования свойств, генерации молекул и поиске ретросинтеза. CX Codex +1 2 100установок boltz-small-molecule-screen официальный Boltz Скилл boltz-small-molecule-screen прогоняет готовую библиотеку SMILES через Boltz: докинг против мишени, ранжирование попаданий и ADME-метрики по каждой молекуле. CX Codex GP ChatGPT +1 ★ 8 052звёзд репо boltz-small-molecule-adme официальный Boltz Скилл boltz-small-molecule-adme оценивает растворимость, проницаемость и липофильность малых молекул по SMILES. Подходит для первичного ADME-отбора без белковой мишени. CX Codex GP ChatGPT +1 ★ 8 052звёзд репо claude-scientific-skills Microck Скилл claude-scientific-skills для Claude Code объединяет 128+ научных скиллов и помогает запускать анализ генома и молекулярный докинг. CX Codex +1 4установок

Из того же набора136

Весь набор →
tooluniverse-toxicology Скилл tooluniverse-toxicology собирает картину токсичности препарата или химиката: механизм по AOP, сигналы FAERS, предупреждения из инструкций FDA и данные EPA CompTox. CX Codex +1 250установок tooluniverse-drug-mechanism-research Скилл tooluniverse-drug-mechanism-research для Claude Code исследует механизм действия лекарства от первичной мишени до клинического исхода. Разбирает фармакологию и сравнивает препараты. CX Codex +1 250установок tooluniverse-regulatory-genomics Скилл tooluniverse-regulatory-genomics ищет регуляторные элементы генома, мотивы связывания факторов транскрипции и оценивает влияние некодирующих мутаций через базы данных и модели. CX Codex +1 247установок tooluniverse-data-integration-analysis Скилл tooluniverse-data-integration-analysis связывает списки генов, мутаций и метаболитов с биологическими базами данных. Помогает понять механизм за статистикой. CX Codex +1 246установок tooluniverse-data-wrangling Скилл tooluniverse-data-wrangling для Claude Code задаёт правила выгрузки и обработки научных данных через Python в Bash. Переключает агента на код для пакетных выгрузок, тысяч строк и редких форматов. CX Codex +1 245установок tooluniverse-protein-structure-prediction Скилл tooluniverse-protein-structure-prediction для Claude Code предсказывает структуру белка по аминокислотной последовательности и собирает данные для оценки. Анализирует новый белок и влияние варианта на структуру. CX Codex +1 245установок tooluniverse-chemical-sourcing Скилл tooluniverse-chemical-sourcing ищет коммерческие источники химических соединений в базах вендоров через PubChem, ChEMBL, ZINC, Enamine, eMolecules и Mcule. CX Codex +1 243установок tooluniverse-clinical-data-integration Скилл tooluniverse-clinical-data-integration для Claude Code собирает обзор безопасности препарата по данным FDA, FAERS, фармакогеномики, исследований и публикаций. Используется в фармаконадзоре и регуляторных оценках. CX Codex +1 242установок

Спонсоры проекта

Компании, которые помогают держать аналитику открытой