Documentation

How to read a HaploReg result

A GWAS hit rarely points at the causal variant itself. It points at a haplotype block: a set of variants that are inherited together, any one of which could be responsible. HaploReg expands each query variant to its LD partners, then lines up the evidence that each candidate is functional.

1. Start with the LD plot. Every point is a variant; height and darkness show its r² with the query (the ringed point). Variants near the top are statistically hard to tell apart from the lead SNP.
2. Look for regulatory context in a relevant tissue. Promoter and enhancer chips (red and amber dots) list the tissues where the variant sits in active chromatin. The Chromatin states tab shows all 127 epigenomes at once; a run of yellow enhancer states in, say, pancreatic islets is a strong hint for a diabetes locus.
3. Check mechanism. A motif change (pink dot) for a factor expressed in that tissue, a bound protein (green) or open chromatin (grey) makes a regulatory effect plausible.
4. Find the target gene. eQTLs link the variant to a gene's expression. The nearest gene is only a first guess.
5. Sort by Evidence to rank candidates by how many independent lines of evidence they have, then open a variant for full detail.

Quick start

Search with one or more rsIDs (rs7903146, rs1421085), a region (chr10:112,990,000-113,000,000), an uploaded text file with one rsID per line, or a curated trait set. Set the LD r² threshold and population; changing them re-runs the query. Display options (epigenome model, conservation, gene annotation) update the table instantly.

Use the filter chips to keep only variants with a given kind of evidence, click column headers to sort, and click any row, plot point or heatmap row to open the variant's record. Export TSV downloads the full annotation table; Copy link shares the exact query and settings.

Prefer the original look? HaploReg Classic uses the same data and settings with the v4 interface.

Column guide

Reference tables

ChromHMM core 15-state model

ChromHMM 25-state model (12 imputed marks)

Tissue codes

Epigenome groups

Text / TSV export columns

The HaploReg v5 pipeline

Version 5 rebuilds the database from the 1000 Genomes high-coverage call set (30×, GRCh38). For each chromosome the pipeline keeps #CHROM, POS, ID, REF, ALT and the per-superpopulation allele frequencies in unrelated samples (AF_*_unrel), assigns rsIDs, annotates every variant with its ChromHMM state in each reference epigenome, and scans the reference and alternate alleles against the motif library with MOODS. Per-chromosome VCFs are bgzipped, tabix-indexed and merged; chromosome Y, which has no ChromHMM annotation, carries the same headers with . values.

Both interfaces read data through one adapter layer, so the pipeline output will replace the interim live source without changes to either UI. See README.md for the variant record format.

What changed from v4

• Annotations computed natively on GRCh38, matching the high-coverage call set (v4 annotated on hg19 and displayed hg38 positions).
• Five superpopulations (AFR, AMR, EAS, EUR, SAS). v4's single Asian group (ASN) held only East Asian samples and corresponds to EAS; South Asian (SAS) populations are new.
• Live data from Ensembl while the v5 backend is built, plus an offline demo mode.
• A redesigned interface adds the LD plot, a chromatin-state heatmap, evidence filters and ranking, a record view for each variant, dark mode and this glossary. The v4 interface remains as Classic.

References

Contact: haploreg@mit.edu