Gene search and selection
Search and select a gene to view its features
Prediction Summary
Predicts whether this promoter will be durably silenced by epigenetic effectors (dCas9–DNMT3A, KRAB, EZH2…) or resist repression. Highest confidence when both methods agree.
How to interpret
- Permissive
- Silencing is expected to stick — chromatin state favors stable repression (Polycomb-repressed or quiescent).
- Refractory
- Silencing is likely transient — active marks (H3K36me3, H3K27ac, Pol II, CTCF) resist repression.
- Unknown
- Chromatin signal is ambiguous or classifier confidence is low.
- Conflicting
- Literature rules return both active and repressive signatures at the same locus (lit rules only).
Literature rules apply 75th/25th percentile thresholds on H3K36me3, H3K27ac, RNA Pol II, CTCF, H3K27me3, and H3K4me3 (O’Geen et al. 2022; Broche et al. 2021).
ChromHMM model is a classifier trained on HEK293 DNMT3A silencing dynamics (Broche et al. 2021) over the 18-state ChromHMM landscape, then applied cross-tissue to liver promoters. Labels: P(permissive) ≥ 0.75 → Permissive, ≤ 0.50 → Refractory, else Unknown.
Select 2+ genes above to compare their features
Select 2+ genes to see a feature heatmap
PCA uses all promoters and selected features (RNA-seq TPM is color-only). Static features are Z-scored; epigenomic, ChromHMM, and other functional signals use log₁₀(x+1) then Z-score.
Interactive view of liver ENCODE bigWigs and canonical promoter annotations. Tracks stream directly from ENCODE's public S3 CDN (range-request capable), so no local data download is required. Select a gene from the search above to navigate to its locus.