Ready to supercharge your AI agents?

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

Tme Immune Profiling Agent · FreedomIntelligence | DecimalAI

Tme Immune Profiling Agent

SkillScore91Maintained

tme-immune-profiling-agent

Comprehensive AI-powered tumor microenvironment immune profiling integrating bulk deconvolution, single-cell analysis, and spatial transcriptomics for immunotherapy biomarker 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/tme-immune-profiling-agent/SKILL.md

decimalai skills pull tme-immune-profiling-agent --out .claude/skills/

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

curl -fsSL --create-dirs -o .claude/skills/tme-immune-profiling-agent/SKILL.md https://app.decimal.ai/s/tme-immune-profiling-agent/SKILL.md

Raw file: /s/tme-immune-profiling-agent/SKILL.md ↗ · machine-readable summary at /s/tme-immune-profiling-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

23 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 — +35% lift · at higher cost+52% tokens0% turns91% of whole cases pass with skill (21/23)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-03	✗→✓	▲ Improved	66%	0%
case-06	✗→✓	▲ Improved	133%	0%
case-10	✗→✓	▲ Improved	82%	0%
case-11	✗→✓	▲ Improved	38%	0%
case-12	✗→✓	▲ Improved	13%	0%

Showing 5 of 23 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/tme-immune-profiling-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

-->

TME Immune Profiling Agent

The TME Immune Profiling Agent provides comprehensive tumor microenvironment (TME) immune profiling by integrating multiple data modalities including bulk RNA-seq deconvolution, single-cell transcriptomics, spatial transcriptomics, and multiplex immunofluorescence. It enables biomarker discovery for immunotherapy response and TME-based patient stratification.

When to Use This Skill

When characterizing immune composition of tumor microenvironment.
For predicting immunotherapy response from TME profiles.
To identify immune cell states and functional programs.
When analyzing spatial organization of immune infiltrates.
For discovering TME-based biomarkers and therapeutic targets.

Core Capabilities

Bulk Deconvolution: Estimate immune cell fractions from bulk RNA-seq.

Single-Cell Immune Profiling: Deep characterization of immune populations.

Spatial Immune Architecture: Map immune cell locations and neighborhoods.

Immune Phenotype Classification: Hot/cold/excluded tumor classification.

Functional State Analysis: Exhaustion, activation, memory signatures.

Response Prediction: Multi-modal immunotherapy response models.

Immune Cell Types Profiled

| Cell Type | Subtypes | Key Markers | |-----------|----------|-------------| | T cells | CD8+, CD4+, Treg, Th1/2/17 | CD3, CD8, CD4, FOXP3 | | B cells | Naive, memory, plasma | CD19, CD20, CD138 | | NK cells | CD56bright, CD56dim | NKG7, NCAM1 | | Macrophages | M1, M2, TAM | CD68, CD163, CD206 | | Dendritic | cDC1, cDC2, pDC | CLEC9A, CD1C, BDCA2 | | MDSC | M-MDSC, PMN-MDSC | CD33, CD11b, ARG1 | | CAF | myCAF, iCAF, apCAF | FAP, ACTA2, COL1A1 |

Deconvolution Methods

| Method | Algorithm | Cell Types | Best For | |--------|-----------|------------|----------| | CIBERSORTx | SVR | 22 | Gold standard | | xCell | ssGSEA | 64 | Comprehensive | | EPIC | Constrained regression | 8 | Tumor/stroma | | MCP-counter | Marker genes | 10 | Robust scores | | quanTIseq | Deconvolution | 10 | Pan-cancer | | TIMER2.0 | Multiple | Variable | Integrated |

Workflow

Input: Bulk RNA-seq, scRNA-seq, spatial data, or IHC images.

Deconvolution: Estimate cell fractions from bulk data.

Single-Cell Analysis: Deep immune phenotyping if available.

Spatial Mapping: Localize immune populations in tissue.

Integration: Combine modalities for comprehensive profile.

Classification: Assign TME phenotype (hot/cold/excluded).

Output: Immune profiles, visualizations, response predictions.

Example Usage

User: "Profile the tumor microenvironment of this lung cancer cohort to identify immunotherapy responders."

Agent Action:

bash
python3 Skills/Immunology_Vaccines/TME_Immune_Profiling_Agent/tme_profiling.py \
    --bulk_rna expression_matrix.tsv \
    --scRNA_data scRNA_lung.h5ad \
    --spatial_data visium_tumor.h5ad \
    --cancer_type nsclc \
    --deconvolution_methods cibersortx,epic,mcpcounter \
    --response_labels clinical_response.csv \
    --output tme_profiles/

TME Phenotypes

| Phenotype | Characteristics | Immunotherapy Response | |-----------|-----------------|----------------------| | Immune Hot | High TIL infiltration, PD-L1+ | Favorable | | Immune Cold | Low TIL, low inflammation | Poor | | Immune Excluded | TILs at margin, not penetrating | Intermediate | | Immune Suppressed | TILs + MDSCs/Tregs | Variable |

Output Components

| Output | Description | Format | |--------|-------------|--------| | Cell Fractions | Per-sample immune estimates | .csv | | TME Classification | Hot/cold/excluded labels | .csv | | Immune Scores | Composite signatures | .csv | | Spatial Maps | Cell type locations | .h5ad | | Neighborhood Analysis | Immune niches | .csv | | Response Prediction | IO probability | .json | | Visualizations | Deconvolution plots | .png, .pdf |

Immune Signatures

| Signature | Genes | Interpretation | |-----------|-------|----------------| | Cytotoxic | PRF1, GZMB, GNLY | T cell killing | | Exhaustion | PDCD1, LAG3, HAVCR2, TIGIT | T cell dysfunction | | IFN-gamma | IFNG, STAT1, IRF1 | Inflammation | | TLS | CD20, CD4, BCL6 | Tertiary lymphoid | | Exclusion | TGFB1, FAP, COL1A1 | Stromal barrier |

AI/ML Components

Deconvolution Enhancement:

Deep learning deconvolution
Multi-method ensemble
Single-cell reference optimization

Response Prediction:

Multi-modal fusion (bulk + spatial)
Survival analysis integration
Transfer learning across cancers

Spatial Analysis:

Graph neural networks for niches
Attention for region importance
Cell-cell interaction networks

Clinical Applications

| Application | TME Feature | Clinical Decision | |-------------|-------------|-------------------| | IO Selection | Immune hot phenotype | Prioritize IO | | Combination | Cold + excluded | Consider combo | | Prognosis | TLS presence | Favorable outcome | | Biomarker | CD8+ density | Response prediction | | Resistance | MDSC enrichment | Address suppression |

Performance Benchmarks

| Task | Dataset | Performance | |------|---------|-------------| | IO Response | NSCLC | AUC 0.78 | | IO Response | Melanoma | AUC 0.82 | | TME Classification | Pan-cancer | Accuracy 85% | | Survival | TCGA | C-index 0.72 |

Prerequisites

Python 3.10+
CIBERSORTx, EPIC, xCell
Scanpy, Squidpy
PyTorch for deep learning
R for certain deconvolution methods

Related Skills

TCR_Repertoire_Analysis_Agent - T cell specificity
TCell_Exhaustion_Analysis_Agent - Exhaustion phenotyping
Spatial_Epigenomics_Agent - Spatial analysis
Nicheformer_Spatial_Agent - Spatial foundation models

Spatial Immune Metrics

| Metric | Definition | Clinical Relevance | |--------|------------|-------------------| | Immune Distance | Distance to tumor edge | Exclusion | | Clustering Coefficient | Immune aggregation | TLS formation | | CD8/Treg Ratio | Spatial ratio | Effector balance | | Contact Score | Immune-tumor contacts | Direct killing | | Neighborhood Entropy | Mixing vs segregation | TME organization |

Special Considerations

Reference Panel: Use cancer-type specific references
Batch Correction: Normalize across samples/platforms
Purity Effects: Account for tumor purity in deconvolution
Single-Cell Validation: Validate bulk estimates with scRNA
Spatial Context: Bulk loses spatial information

Therapeutic Implications

| TME State | Therapeutic Strategy | |-----------|---------------------| | Hot, PD-L1+ | Anti-PD-1/PD-L1 | | Cold | Oncolytic virus, radiation, chemo | | Excluded | TGF-beta inhibition, VEGF targeting | | Suppressed | Treg depletion, MDSC targeting | | TLS+ | Excellent IO candidate |

Author

AI Group - Biomedical AI Platform

Illustrative Want to see what tme-immune-profiling-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, 18:22 UTC · 23 cases

Lift

+35%

91% vs 57%

Agent turns

1.0 vs 1.0

Tokens

+52%

4,240 vs 2,782.87

fail → pass · 8pass → pass · 13fail → fail · 2

Per-case results (23)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 I have bulk RNA-seq transcriptomic profiles and clinical response annotations from a non-small cell lung cancer cohort. Please analyze the immune cell composition across these samples, assign tumor microenvironment phenotype classifications, and estimate immunotherapy response likelihoods. Return the results as CSV files for cell fractions and phenotype categories, a JSON summary for response predictions, and standard PNG visualization charts.	fail→fail	27,827	27,726	-0%	1	1	0%	6,214	8,126	+31%	0	0	—
case-02 We possess matched single-cell RNA-seq and spatial Visium transcriptomic datasets for melanoma tissue biopsies. We need a spatial immune analysis that calculates cell-cell proximity measures, distance metrics to the tumor border, and cellular neighborhood structures. Provide the output as an updated h5ad object containing spatial mappings, a CSV matrix of spatial immune metrics, and PDF summary plots of the spatial niches.	fail→fail	23,552	25,283	+7%	1	1	0%	5,227	7,140	+37%	0	0	—
case-03 A researcher wants to choose deconvolution algorithms for bulk RNA-seq. They specifically need one method suited for constrained regression to separate tumor vs stroma across 8 cell types and another SVR-based method for 22 detailed immune subsets. Which methods fit these exact specifications?	fail→pass	12,645	4,577	-64%	1	1	0%	1,659	2,747	+66%	0	0	—
case-04 We are building a multi-color immunofluorescence panel for stromal barriers and myeloid suppression in triple-negative breast cancer. Which specific marker combinations characterize CAFs and MDSCs in this profiling paradigm?	pass→pass	17,441	19,733	+13%	1	1	0%	3,260	5,221	+60%	0	0	—
case-05 An analyst is constructing gene set scoring modules for immune functional states in tumor transcriptomics. What specific gene sets define the T cell exhaustion signature versus the stromal exclusion signature?	pass→pass	16,482	12,515	-24%	1	1	0%	3,069	4,219	+37%	0	0	—
case-06 When evaluating spatial interactions between CD8+ T cells and cytokeratin-positive tumor cells in multiplex imaging, which metric measures direct immune-tumor cell physical contacts versus the degree of spatial mixing across the microenvironment?	fail→pass	10,165	10,730	+6%	1	1	0%	1,614	3,765	+133%	0	0	—
case-07 Biopsy evaluation of a metastatic renal cell carcinoma specimen reveals high numbers of CD8+ T cells locked in the dense peritumoral stroma, unable to infiltrate the core. What therapeutic target combination should be prioritized based on this TME state?	pass→pass	9,952	9,509	-4%	1	1	0%	1,803	3,404	+89%	0	0	—
case-08 Bulk transcriptomics and flow cytometry show dense CD8 T cell infiltration accompanied by high levels of FOXP3+ Tregs and ARG1+ PMN-MDSCs. How should the therapeutic strategy be tailored for this specific immunosuppressive microenvironment?	pass→pass	16,273	16,572	+2%	1	1	0%	3,006	4,904	+63%	0	0	—
case-09 In single-cell RNA-seq annotation of tumor-infiltrating myeloid cells, what marker set differentiates cDC1, cDC2, and pDC populations?	pass→pass	13,653	14,803	+8%	1	1	0%	2,621	4,371	+67%	0	0	—
case-10 We are implementing a spatial transcriptomics pipeline and need exact definitions for assessing cell proximity relative to the invasive front versus local cellular grouping. What spatial metrics correspond to distance to tumor edge and immune aggregation?	fail→pass	16,927	19,824	+17%	1	1	0%	3,207	5,821	+82%	0	0	—
case-11 A clinical trial team wants to stratify patients by the presence of ectopic lymphoid structures in lung adenocarcinoma samples. What core marker gene set defines the TLS signature?	fail→pass	13,745	8,723	-37%	1	1	0%	2,562	3,525	+38%	0	0	—
case-12 To evaluate baseline inflammatory signaling in bulk RNA-seq samples prior to checkpoint blockade, which gene trio constitutes the IFN-gamma signature module?	fail→pass	11,867	3,076	-74%	1	1	0%	2,213	2,496	+13%	0	0	—
case-13 A data scientist wants to compare xCell and MCP-counter for bulk RNA-seq processing. What underlying algorithm and cell type capacity does xCell use versus MCP-counter?	pass→pass	12,617	9,286	-26%	1	1	0%	2,263	3,234	+43%	0	0	—
case-14 Write the command-line invocation for running the TME profiling CLI on bulk RNA (`bulk.tsv`), scRNA (`sc.h5ad`), spatial Visium (`spatial.h5ad`), for melanoma with CIBERSORTx and EPIC deconvolution, saving to `results/`.	fail→pass	7,134	2,427	-66%	1	1	0%	1,358	2,409	+77%	0	0	—
case-15 When profiling tumor-associated macrophages in human solid tumors, what key surface and endocytic marker genes distinguish general, M1, and M2 TAM states?	pass→pass	16,808	19,078	+14%	1	1	0%	3,257	5,131	+58%	0	0	—
case-16 Single-cell analysis of tumor-infiltrating lymphocytes reveals NK cell populations. What marker genes confirm NK cell lineage in this matrix?	fail→pass	13,313	10,920	-18%	1	1	0%	2,607	3,980	+53%	0	0	—
case-17 When deploying multi-modal response prediction models on clinical cohorts, what area under the ROC curve performance benchmark is established for predicting immunotherapy response in melanoma?	pass→pass	11,556	2,284	-80%	1	1	0%	1,964	2,272	+16%	0	0	—
case-18 What classification accuracy benchmark is established when stratifying pan-cancer tissue samples into TME phenotypes?	pass→pass	12,729	1,510	-88%	1	1	0%	2,277	2,125	-7%	0	0	—
case-19 What special methodological considerations must be addressed when performing bulk RNA-seq deconvolution on heterogeneous solid tumor biopsies to prevent inaccurate cell fraction estimates?	fail→pass	18,251	11,575	-37%	1	1	0%	3,045	3,949	+30%	0	0	—
case-20 A gastric cancer specimen shows high TIL infiltration and strong PD-L1 positivity throughout the core. How is this phenotype classified and what is its expected response to single-agent immune checkpoint inhibition?	pass→pass	7,767	8,998	+16%	1	1	0%	1,479	3,249	+120%	0	0	—
case-21 We have bulk V(D)J T-cell receptor sequencing data from peripheral blood of melanoma patients. We want to align reads to IMGT references, calculate Shannon entropy of CDR3 length distributions, and identify expanded clonotypes. Please outline the pipeline for TCR V(D)J processing.	pass→pass	15,449	20,191	+31%	1	1	0%	2,682	5,772	+115%	0	0	—
case-22 I have raw single-cell RNA-seq count matrices from peripheral blood mononuclear cells (PBMCs) of healthy human donors. I need to run standard Seurat v5 quality control filtering, log-normalization, PCA dimension reduction, Louvain clustering, and UMAP visualization. Provide the standard R/Seurat script.	pass→pass	12,844	11,885	-7%	1	1	0%	2,638	4,498	+71%	0	0	—
case-23 We collected bulk ATAC-seq fastq files from isolated CD8+ memory T cells. We need to align reads to hg38 using Bowtie2, call open chromatin peaks with MACS3, and perform TF motif enrichment analysis with HOMER. Provide the shell workflow.	pass→pass	18,468	16,288	-12%	1	1	0%	3,981	5,162	+30%	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. 23 cases were attempted. The headline lift of +35 percentage points is the difference between those two pass rates over the 23 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 56.5%
Pass rate with the skill: 91.3%
Cases attempted: 23
Cases counted toward lift: 23
Cases that regressed: 0
Skill version measured: v2
Run completed: 8/2/2026, 18:22 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-03

Prompt

A researcher wants to choose deconvolution algorithms for bulk RNA-seq. They specifically need one method suited for constrained regression to separate tumor vs stroma across 8 cell types and another SVR-based method for 22 detailed immune subsets. Which methods fit these exact specifications?

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 run91

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+56% 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	+17%

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