OmniBioAI

OmniLink: Enhancer-to-Target Mapping

Map non-coding variants to enhancer peaks and chromatin loops to infer target genes. Includes ABC-like scoring (ATAC + Hi-C + distance penalty).

Category: genomics Engine: local_python Output: variant_gene.tsv

Refresh

Provide variants, ATAC peaks, and Hi-C loops. (GTF is recommended for real promoter overlap.)

Variants (TSV or BED) *

TSV recommended: chrom, pos (1-based), optional id, pval.

ATAC peaks (BED / narrowPeak) *

Uses column 5 (score) as peak_signal when present.

Hi-C loops (BEDPE) *

Expected: chr1 start1 end1 chr2 start2 end2 [score].

Gene annotation (GTF) (optional)

If provided, OmniLink computes promoter_overlap=True for true promoter matches.

Genome build

Used as a label for now (no liftover in MVP).

Promoter window (bp)

Promoter region = TSS ± window.

Max loop distance (bp)

Loops beyond this span are ignored (same-chrom).

Top N variant→gene links

Limits output size for ranking table.

Min loop score (optional)

Filter loops: keep only those with score ≥ threshold.

Interpretable formula: (log1p(peak)+log1p(loop)+1) * distance_weight * promoter_weight

Tune distance penalty + promoter boost.

Tip: start with defaults. Increase abc_power if distant loops dominate; increase abc_peak_scale if ATAC activity should weigh more.

Outputs will appear on the run details page.