Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Pan Cancer Multiomics Agent · FreedomIntelligence | DecimalAI

Pan Cancer Multiomics Agent

SkillScore86Maintained

pan-cancer-multiomics-agent

AI-powered pan-cancer analysis integrating genomic, transcriptomic, proteomic, and epigenomic data for cancer subtyping, driver identification, and cross-cancer pattern discovery.

scienceby FreedomIntelligence

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Research pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/pan-cancer-multiomics-agent/SKILL.md

decimalai skills pull pan-cancer-multiomics-agent --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/pan-cancer-multiomics-agent/SKILL.md https://app.decimal.ai/s/pan-cancer-multiomics-agent/SKILL.md

Raw file: /s/pan-cancer-multiomics-agent/SKILL.md ↗ · machine-readable summary at /s/pan-cancer-multiomics-agent.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 51d ago

Yes — +45% lift · at higher cost+35% tokens0% turns86% of whole cases pass with skill (19/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-02	✗→✓	▲ Improved	39%	0%
case-09	✗→✓	▲ Improved	108%	0%
case-10	✗→✓	▲ Improved	30%	0%
case-11	✗→✓	▲ Improved	18%	0%
case-12	✗→✓	▲ Improved	-27%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/1/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

FreedomIntelligence / OpenClaw-Medical-Skills

skills/pan-cancer-multiomics-agent/SKILL.md

View on GitHub ↗

Imported 8/1/2026Version v2

Skill contents

SKILL.md

<!--

COPYRIGHT NOTICE

This file is part of the "Universal Biomedical Skills" project.

Copyright (c) 2026 MD BABU MIA, PhD <md.babu.mia@mssm.edu>

All Rights Reserved.

This code is proprietary and confidential.

Unauthorized copying of this file, via any medium is strictly prohibited.

Provenance: Authenticated by MD BABU MIA

-->

Pan-Cancer Multi-Omics Agent

The Pan-Cancer Multi-Omics Agent integrates multi-omics data across cancer types to identify shared oncogenic drivers, discover novel subtypes, and enable cross-cancer therapeutic insights. It leverages TCGA, CPTAC, and other pan-cancer resources with deep learning for comprehensive cancer characterization.

When to Use This Skill

When analyzing patient tumors in context of pan-cancer molecular profiles.
To identify shared drivers and vulnerabilities across cancer types.
For discovering novel molecular subtypes that span histological boundaries.
When prioritizing therapeutic targets with pan-cancer evidence.
To benchmark single-cancer findings against pan-cancer patterns.

Core Capabilities

Pan-Cancer Subtyping: ML-based clustering across 32+ cancer types to identify molecular subtypes transcending tissue of origin.

Driver Discovery: Integrate mutation, expression, and CNV data to identify oncogenic drivers using pan-cancer statistical power.

Multi-Omics Fusion: Deep learning integration of mRNA, miRNA, methylation, and protein data for comprehensive profiles.

Pathway Analysis: Identify dysregulated pathways with pan-cancer prevalence and therapeutic implications.

Survival Modeling: PRISM framework for multi-omics prognostic marker discovery and survival prediction.

Therapeutic Matching: Map patient profiles to pan-cancer drug sensitivity data and clinical trial evidence.

TCGA Pan-Cancer Atlas Integration

| Data Type | Samples | Application | |-----------|---------|-------------| | Somatic mutations | 11,000+ | Driver identification | | Copy number | 11,000+ | Amplifications/deletions | | mRNA expression | 11,000+ | Expression subtypes | | miRNA expression | 10,000+ | Regulatory networks | | DNA methylation | 10,000+ | Epigenetic subtypes | | Protein (RPPA) | 8,000+ | Pathway activation |

Workflow

Input: Patient multi-omics data (mutations, CNV, expression, methylation).

Normalization: Harmonize data to TCGA reference standards.

Classification: Assign to pan-cancer molecular subtypes.

Driver Analysis: Identify patient-specific drivers in pan-cancer context.

Pathway Scoring: Calculate pathway activation scores.

Therapeutic Matching: Identify actionable targets and trial matches.

Output: Pan-cancer classification, driver report, pathway profiles, treatment recommendations.

Example Usage

User: "Classify this breast cancer patient's tumor in the pan-cancer context and identify shared drivers."

Agent Action:

bash
python3 Skills/Oncology/Pan_Cancer_MultiOmics_Agent/pancancer_analyzer.py \
    --mutations patient_mutations.maf \
    --expression patient_rnaseq.tsv \
    --methylation patient_methylation.tsv \
    --cnv patient_cnv_segments.tsv \
    --reference tcga_pancancer \
    --subtype_method nmf_consensus \
    --output pancancer_report/

Pan-Cancer Molecular Subtypes

Cross-cancer molecular taxonomy identifies patterns beyond histology:

| Subtype | Characteristics | Example Cancers | |---------|-----------------|-----------------| | C1-Wound healing | High proliferation, MYC amp | Breast, ovarian, bladder | | C2-IFN-gamma dominant | Immune active, high TCR/BCR | Melanoma, lung, cervical | | C3-Inflammatory | NF-kB, cytokine signatures | Head/neck, stomach | | C4-Lymphocyte depleted | Low immune, PTEN loss | Glioma, uveal melanoma | | C5-Immunologically quiet | Low expression overall | Kidney chromophobe, thyroid | | C6-TGF-beta dominant | High TGF-B, fibrosis | Pancreas, rectum, glioma |

Deep Learning Architecture

Multi-Omics Integration Model:

Input Layers:
  - Genomic encoder (mutations, CNV)
  - Transcriptomic encoder (mRNA, miRNA)
  - Epigenomic encoder (methylation)
  - Proteomic encoder (RPPA)

Fusion Layer:
  - Cross-attention mechanism
  - Multi-modal variational autoencoder

Output Heads:
  - Subtype classifier
  - Survival predictor
  - Drug response predictor

MLOmics Database Access

The agent integrates with MLOmics, providing:

8,314 patient samples across 32 cancer types
Pre-computed features for ML benchmarking
Standardized train/test splits for reproducibility
Drug sensitivity data for 300+ compounds

Prerequisites

Python 3.10+
PyTorch with multi-modal architectures
Access to TCGA, CPTAC, or local data
16GB+ RAM for pan-cancer analysis

Related Skills

Tumor_Clonal_Evolution - For intratumoral heterogeneity
Multi_Omics_Integration - For single-patient integration
Drug_Repurposing - For therapeutic matching

Clinical Applications

Cancer of Unknown Primary (CUP): Identify tissue of origin
Cross-indication trials: Find basket trial eligibility
Driver prioritization: Pan-cancer functional evidence
Prognosis: Multi-omics survival models

Author

AI Group - Biomedical AI Platform

Illustrative Want to see what pan-cancer-multiomics-agent would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/2/2026, 18:04 UTC · 22 cases

Lift

+45%

86% vs 41%

Agent turns

1.0 vs 1.0

Tokens

+35%

3,242.41 vs 2,406.55

fail → pass · 10pass → pass · 9fail → fail · 3

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-04 We performed single-cell DNA and RNA sequencing on multiple biopsies taken from a primary glioblastoma and its recurrent relapse. Can you construct a phylogenetic tree of subclonal evolution and reconstruct the longitudinal clonal lineage hierarchy across tumor progression?	pass→pass	16,323	27,854	+71%	1	1	0%	2,834	6,586	+132%	0	0	—
case-05 We solved the 2.1 Angstrom crystal structure of a mutant kinase binding pocket. Can you run virtual screening and molecular docking simulations against a 50,000 compound chemical library to predict binding affinity constants (Kd)?	pass→pass	14,498	21,952	+51%	1	1	0%	2,689	5,397	+101%	0	0	—
case-06 We have RNA-seq transcript counts from pre-treatment and post-treatment blood samples from a single melanoma patient receiving targeted therapy. Can you perform standard DESeq2 differential gene expression analysis between pre- and post-treatment without cross-referencing pan-cancer cohorts?	pass→pass	12,552	14,498	+16%	1	1	0%	2,240	4,100	+83%	0	0	—
case-07 A patient tumor profile reveals intense cellular proliferation, microenvironment alteration, and MYC locus amplification across breast and ovarian tissue samples. How is this specific molecular profile classified within cross-cancer taxonomy, and what are its key features?	pass→pass	13,638	7,452	-45%	1	1	0%	2,429	2,530	+4%	0	0	—
case-01 I have sequencing data from a patient tumor including MAF mutations, RNA-seq expression, CNV segments, and DNA methylation data. Please harmonize these files against reference standards to assign the sample to a pan-cancer consensus molecular subtype, evaluate driver mutations, and provide a report on activated pathways and potential therapeutic matches.	fail→fail	17,437	27,819	+60%	1	1	0%	2,900	7,168	+147%	0	0	—
case-02 We just generated multi-modal profiling files (somatic mutation list, gene expression matrix, and copy number variations) for a high-grade breast carcinoma specimen. Can you cross-reference these omics layers against pan-cancer reference cohorts to rank the key functional drivers, assess survival risk, and generate a list of prioritized drug sensitivities and basket trial matches?	fail→pass	26,219	26,682	+2%	1	1	0%	4,393	6,119	+39%	0	0	—
case-03 Please take our multi-omics patient dataset containing somatic mutation calls, copy number alterations, methylation beta values, and transcriptomic counts. Integrate these datasets across genomic, epigenomic, and transcriptomic dimensions to determine its cross-cancer molecular subtype classification, compute dysregulated pathway scores, and output a prioritized report of oncogenic drivers.	fail→fail	23,665	30,786	+30%	1	1	0%	4,824	7,547	+56%	0	0	—
case-08 A melanoma tumor sample demonstrates high immune infiltration, elevated T-cell receptor (TCR) and B-cell receptor (BCR) diversity, and strong interferon-gamma signature signaling. Which pan-cancer immune subtype classification applies to this tumor?	pass→pass	4,440	4,500	+1%	1	1	0%	772	2,114	+174%	0	0	—
case-09 A high-grade glioma specimen displays low overall immune cell infiltration, marked loss of PTEN, and poor prognosis. Which pan-cancer molecular subtype best categorizes this tumor state?	fail→pass	5,713	3,362	-41%	1	1	0%	933	1,938	+108%	0	0	—
case-10 Compare the representative cancer types associated with the C5 Immunologically quiet and C6 TGF-beta dominant pan-cancer subtypes.	fail→pass	14,192	11,518	-19%	1	1	0%	2,474	3,222	+30%	0	0	—
case-11 We need to execute the standard pan-cancer multi-omics analyzer script on a cohort using patient mutations (in MAF format), RNA-seq counts (TSV format), DNA methylation (TSV format), and copy-number segment files (TSV format), setting the reference cohort to TCGA pan-cancer and subtyping method to consensus NMF. What exact CLI command invocation performs this task?	fail→pass	9,945	3,733	-62%	1	1	0%	1,807	2,131	+18%	0	0	—
case-12 When configuring statistical power thresholds for multi-omics integration against the TCGA Pan-Cancer Atlas reference cohort, what are the sample size capacities available across somatic mutations, protein RPPA, and miRNA data?	fail→pass	15,037	4,894	-67%	1	1	0%	2,631	1,923	-27%	0	0	—
case-13 We are setting up a benchmark for cross-cancer machine learning models using the integrated MLOmics database. What are the sample size, cancer type count, and compound sensitivity specifications provided by this resource?	fail→pass	7,257	2,905	-60%	1	1	0%	1,269	1,837	+45%	0	0	—
case-14 Which specialized framework is utilized by the pan-cancer multi-omics agent for prognostic marker discovery and multi-omics survival prediction across cancer cohorts?	fail→pass	11,840	1,496	-87%	1	1	0%	2,063	1,548	-25%	0	0	—
case-15 In the pan-cancer deep learning multi-omics integration model architecture, what structural mechanisms are defined in the fusion layer to merge inputs from the various encoders?	fail→pass	18,282	4,037	-78%	1	1	0%	3,167	2,036	-36%	0	0	—
case-16 In the multi-omics deep learning architecture, which specific omics data types are mapped to the genomic encoder and transcriptomic encoder respectively?	pass→pass	7,370	1,737	-76%	1	1	0%	1,394	1,642	+18%	0	0	—
case-17 What are the three specific prediction output heads constructed in the multi-omics deep learning model architecture?	pass→pass	13,775	1,421	-90%	1	1	0%	2,228	1,522	-32%	0	0	—
case-18 A biopsy from a metastatic lesion of unknown primary origin underwent whole-exome and transcriptomic sequencing. How does the pan-cancer agent resolve the primary tissue and therapeutic options for this tumor?	pass→pass	16,656	12,902	-23%	1	1	0%	3,020	3,753	+24%	0	0	—
case-19 What immunogenomic signature characteristics and signaling pathways define the C3 Inflammatory pan-cancer subtype, and which primary cancer sites frequently manifest this subtype?	fail→pass	12,941	10,171	-21%	1	1	0%	2,241	3,090	+38%	0	0	—
case-20 What are the minimum RAM requirements and Python runtime version specified for executing full pan-cancer multi-omics reference cohort analysis?	fail→fail	28,717	1,762	-94%	1	1	0%	2,486	1,568	-37%	0	0	—
case-21 During the multi-omics workflow, before assigning a patient sample to cross-cancer molecular subtypes, what harmonization step must be performed on raw input matrices?	fail→pass	9,149	2,912	-68%	1	1	0%	1,565	1,724	+10%	0	0	—
case-22 If an analysis requires evaluating single-patient multi-omics integration without cross-cancer cohort benchmarking, or investigating intratumoral heterogeneity and clone structures, which alternative specialized skills should be recommended?	pass→pass	15,523	2,976	-81%	1	1	0%	2,585	1,838	-29%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +45 percentage points is the difference between those two pass rates over the 22 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 40.9%
Pass rate with the skill: 86.4%
Cases attempted: 22
Cases counted toward lift: 22
Cases that regressed: 0
Skill version measured: v2
Run completed: 8/2/2026, 18:04 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-02

Prompt

We just generated multi-modal profiling files (somatic mutation list, gene expression matrix, and copy number variations) for a high-grade breast carcinoma specimen. Can you cross-reference these omics layers against pan-cancer reference cohorts to rank the key functional drivers, assess survival risk, and generate a list of prioritized drug sensitivities and basket trial matches?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run86

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+46% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 51 days ago

Model	Method	Date	Lift
gemini-3.6-flash	verified	7/26/2026	—

Other measured skills in the registry, with their headline benchmark lift.