Help and user guide

The search box on the home page accepts any of these:

• Variant ID in chrom_pos_ref_alt format based on the GRCh38 reference genome, e.g. chr20_54107677_C_A.
• rsID with real-time resolution to GRCh38 coordinates, allowing lookups for variants not explicitly stored in the database, e.g. rs6013892.
• Gene symbol listing scored variants within the gene body plus 100 kb flanking regions, e.g. CYP3A4.
• Region defining a genomic interval in chrom:start-end format, e.g. chr20:54107000-54108000.

Each variant carries an independent effect score for each receptor, shown two ways:

• Raw Δ: the center bin effect, the sum of ALT minus REF on the receptor head. Negative means predicted loss of binding (LoF), positive means gain (GoF).
• Normalized [−1, 1]: Δ scaled to the receptor peak population 99th percentile and clipped. A value near ±1 sits at or above the 99th percentile of that receptor.

On gene tables you can switch raw Δ and normalized with the toggle above the table. The variant page also reports the impact band and, where known, the regulated gene.

All substitutions: every possible base change at a position is scored, so you can see whether the effect is specific to one substitution or whether any change disrupts binding. Observational databases cannot show this.

What a variant page shows:

AhR is exploratory: its head was trained on a single replicate, so AhR scores are reported with that caveat.

• The atlas covers the PXR, FXR and AhR regulatory elements (top ChIP-seq peak summits by signal), not the whole genome. A position outside every receptor element is not scored.
• Saturation variants are every possible SNV inside those elements; observed variants are the subset seen in gnomAD and carry a population allele frequency.
• Nearest gene is not the regulated gene. For a regulatory variant the experimentally linked target (eQTL or curated literature) is shown separately from the nearest gene by distance.
• Scores are research predictions, not clinical advice.

Where available, a variant is cross-annotated with GTEx liver eQTL target, ClinVar record, GWAS Catalog trait, FIMO TF-motif overlap, phyloP conservation, and dbSNP rsID. These appear only when present, so an empty field means no annotation, not an error.

A read only REST API is described on the API page, with variant, gene, region, element, batch, cross receptor and stats endpoints. The full atlas and tables are archived on Zenodo for bulk download.

Please cite the Zenodo record (DOI 10.5281/zenodo.20753827) and see Contact for authorship and data provenance.