v4 primary ranking (rank-percentile sum): Top-1000 per cell type ranked by
v4 strength (plain, no Poisson):
Distal filter + taxonomy separation (unchanged from v3): |dist_to_tss| ≥ 2 kb applied before top-N selection. Primary tabs use log2FC vs the 10 primary lineages; subtype tabs use log2FC vs 13 lineages (pooled malignant excluded) to avoid self-competition.
Amplicon/CN caveat (permanent). Peaks within ±1 Mb of chr7p11.2 EGFR, chr12q15 MDM2, chr8q24 MYC, or chr9p21.3 CDKN2A carry an AMP badge. In NON-malignant tabs the badge flags ambient-fragment contamination from malignant cells. In MALIGNANT tabs the badge flags copy-number inflation of per-cell fragment counts (recurrent amp/del driving CPM, not regulatory activity). No automatic score correction is applied — treat as de-selection hint.
Nearest-gene is informational only. AAV scaffold efficacy depends on cell-type-selective driving strength (what primary_score ranks), not nearest-gene identity.
Hover any column header for its definition and formula.
primary_score = (spec_pct + strength_pct) / 2, where spec_pct is the percentile rank of
log2FC_top_vs_next and strength_pct is the percentile rank of
strength_frags_per_cell_wn, both computed within distal peaks.
v4 strength (plain, no Poisson):
strength = mean over evaluable patients of (fragments / cells × 500 / peak_width_bp).
Just the average number of Tn5 fragments per cell at this peak, width-normalized to a 500 bp reference. No model assumption.
(v3 used 1 - exp(-fragments/cells), which was based on a Poisson assumption that doesn't match diploid biology; dropped in v4.)
Distal filter + taxonomy separation (unchanged from v3): |dist_to_tss| ≥ 2 kb applied before top-N selection. Primary tabs use log2FC vs the 10 primary lineages; subtype tabs use log2FC vs 13 lineages (pooled malignant excluded) to avoid self-competition.
Amplicon/CN caveat (permanent). Peaks within ±1 Mb of chr7p11.2 EGFR, chr12q15 MDM2, chr8q24 MYC, or chr9p21.3 CDKN2A carry an AMP badge. In NON-malignant tabs the badge flags ambient-fragment contamination from malignant cells. In MALIGNANT tabs the badge flags copy-number inflation of per-cell fragment counts (recurrent amp/del driving CPM, not regulatory activity). No automatic score correction is applied — treat as de-selection hint.
Nearest-gene is informational only. AAV scaffold efficacy depends on cell-type-selective driving strength (what primary_score ranks), not nearest-gene identity.
Hover any column header for its definition and formula.
| peak_id | chrom | start | end | width (bp) | nearest gene | dist to TSS | band | log2FC top vs next | spec pct | strength (frags/cell, wn) | strength pct | primary score | n cohorts | patient detection |
|---|