Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →AI-powered spatial epigenomics analysis combining chromatin accessibility, histone modifications, and DNA methylation with spatial coordinates for tissue architecture mapping.
.claude/skills/spatial-epigenomics-agent/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 47% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -17% | 0% |
| case-12 | ✓→✗ | ▼ Worse | 41% | 0% |
<!--
#
#
-->
The Spatial Epigenomics Agent analyzes spatial epigenomic data combining chromatin accessibility (ATAC-seq), histone modifications (CUT&Tag), and DNA methylation with spatial coordinates. It maps regulatory landscapes across tissue architecture to understand cell-state regulation in spatial context.
| Technology | Epigenetic Mark | Resolution | Method | |------------|-----------------|------------|--------| | Spatial-ATAC-seq | Open chromatin | ~10-50μm | Microfluidic barcoding | | DBiT-seq | ATAC + expression | ~10μm | Deterministic barcoding | | Spatial-CUT&Tag | Histone marks | ~50μm | Cleavage under targets | | Spatial-MethylSeq | DNA methylation | Variable | Bisulfite conversion | | MERFISH + epigenetics | 3D organization | Single-cell | Imaging-based |
User: "Analyze this spatial ATAC-seq dataset to identify spatially-variable regulatory elements in the tumor microenvironment."
Agent Action:
bashpython3 Skills/Genomics/Spatial_Epigenomics_Agent/spatial_epigenomics.py \ --input spatial_atac_fragments.tsv.gz \ --coordinates spot_coordinates.csv \ --peaks macs2_peaks.bed \ --spatial_variable true \ --motif_db jaspar_2024 \ --integrate_with spatial_rna.h5ad \ --output spatial_epi_results/
1. Spatial Peak Calling
2. Spatial Domain Detection
3. Transcription Factor Analysis
4. Enhancer-Gene Linking
Spatial ATAC-seq Spatial RNA-seq
| |
v v
Peak Matrix Expression Matrix
| |
+--------> Integration <-+
|
v
Regulatory Network
(Enhancer -> TF -> Gene)| Metric | Description | Typical Range | |--------|-------------|---------------| | TSS Enrichment | Signal at transcription start sites | >4 good quality | | FRiP | Fraction reads in peaks | >30% | | Spatial autocorrelation | Moran's I for epigenetic features | 0.2-0.8 | | Spots per gene | Detection sensitivity | 100-500 |
AI Group - Biomedical AI Platform
<!-- AUTHOR_SIGNATURE: 9a7f3c2e-MD-BABU-MIA-2026-MSSM-SECURE -->
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 29,821 | 29,582 | -1% | 1 | 1 | 0% | 6,227 | 6,856 | +10% | 0 | 0 | — |
case-02 | fail→pass | 12,053 | 3,767 | -69% | 1 | 1 | 0% | 2,122 | 2,153 | +1% | 0 | 0 | — |
case-03 | pass→pass | 14,063 | 11,018 | -22% | 1 | 1 | 0% | 2,370 | 3,064 | +29% | 0 | 0 | — |
case-04 | pass→pass | 9,365 | 7,340 | -22% | 1 | 1 | 0% | 1,598 | 2,548 | +59% | 0 | 0 | — |
case-05 | fail→pass | 12,026 | 10,264 | -15% | 1 | 1 | 0% | 2,095 | 3,079 | +47% | 0 | 0 | — |
case-06 | pass→pass | 6,611 | 6,035 | -9% | 1 | 1 | 0% | 1,209 | 2,407 | +99% | 0 | 0 | — |
case-07 | pass→pass | 5,386 | 5,602 | +4% | 1 | 1 | 0% | 945 | 2,346 | +148% | 0 | 0 | — |
case-08 | pass→pass | 13,516 | 11,746 | -13% | 1 | 1 | 0% | 2,153 | 3,368 | +56% | 0 | 0 | — |
case-09 | pass→pass | 16,180 | 7,966 | -51% | 1 | 1 | 0% | 2,369 | 2,617 | +10% | 0 | 0 | — |
case-10 | pass→pass | 6,677 | 3,964 | -41% | 1 | 1 | 0% | 1,240 | 1,930 | +56% | 0 | 0 | — |
case-11 | pass→pass | 13,831 | 7,220 | -48% | 1 | 1 | 0% | 2,301 | 2,505 | +9% | 0 | 0 | — |
case-12 | pass→fail | 9,160 | 6,044 | -34% | 1 | 1 | 0% | 1,656 | 2,328 | +41% | 0 | 0 | — |
case-13 | pass→pass | 5,518 | 3,704 | -33% | 1 | 1 | 0% | 835 | 1,939 | +132% | 0 | 0 | — |
case-14 | pass→pass | 9,643 | 6,214 | -36% | 1 | 1 | 0% | 1,628 | 2,300 | +41% | 0 | 0 | — |
case-15 | pass→pass | 8,490 | 5,001 | -41% | 1 | 1 | 0% | 1,379 | 2,161 | +57% | 0 | 0 | — |
case-16 | pass→pass | 12,071 | 5,627 | -53% | 1 | 1 | 0% | 2,036 | 2,252 | +11% | 0 | 0 | — |
case-17 | fail→pass | 15,129 | 1,314 | -91% | 1 | 1 | 0% | 2,597 | 1,495 | -42% | 0 | 0 | — |
case-18 | fail→pass | 11,815 | 2,962 | -75% | 1 | 1 | 0% | 2,144 | 1,783 | -17% | 0 | 0 | — |
case-19 | pass→pass | 16,850 | 13,517 | -20% | 1 | 1 | 0% | 2,487 | 3,483 | +40% | 0 | 0 | — |
case-20 | pass→pass | 17,005 | 16,999 | -0% | 1 | 1 | 0% | 3,291 | 3,999 | +22% | 0 | 0 | — |
case-21 | pass→pass | 16,507 | 13,575 | -18% | 1 | 1 | 0% | 3,032 | 4,003 | +32% | 0 | 0 | — |
case-22 | pass→pass | 11,609 | 11,091 | -4% | 1 | 1 | 0% | 2,311 | 3,501 | +51% | 0 | 0 | — |
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 +14 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 7/26/2026 | +9% |
Other measured skills in the registry, with their headline benchmark lift.