/genomics/ › chip-PRS explorer

Chip-PRS Explorer

Drop your 23andMe v5 raw download. See how many of your typed SNPs land in the chr 6 MHC region, and how that overlap compares against European and East Asian schizophrenia GWAS hits. Companion to the methods paper and the EAS landscape viz.

Privacy: Your genotype file is parsed entirely in the browser. It never leaves your device — no upload, no fetch, no analytics event with file content. View source / Network tab to verify. Close the page and the data is gone.

1. Drop your 23andMe v5 file

2. Your chip's coverage

—
Biallelic SNPs typed
—
In chr 6 MHC band
—
% of chip in MHC

The MHC band is chr 6: 25–34 Mb (GRCh37 extended-MHC). About 0.3% of the genome by length sits in this band; consumer chips typically over-sample it for HLA imputation, so the chip-coverage fraction tends to run higher.

3. Overlap with schizophrenia GWAS hits

Cohort & partition GWS SNPs Typed by your chip Coverage
EUR (PGC3 wave 3) — all———
  in chr 6 MHC———
EAS (Lam 2019) — all———
  in chr 6 MHC———

4. Reference: Dave's chip

For comparison, the page author's own 23andMe v5 chip, computed against the same GWS sets this tool uses:

  • 594,952 biallelic SNPs typed (94.1% of 631,970 raw)
  • 9,074 typed SNPs land in the chr 6 MHC band (1.53% of chip; about 5x the genome-fraction baseline of ~0.3%)
  • 1,363 overlap PGC3 wave 3 EUR GWS — 798 (58.5%) in MHC, 565 (41.5%) outside
  • 54 overlap Lam 2019 EAS GWS — 0 (0%) in MHC, 54 (100%) outside

Note: the companion methods paper reports slightly different EUR overlap numbers (1,491 / 990) because it uses a smaller EUR-only PGC3 release variant (52,017 cases) rather than the larger primary release used here (71,554 cases). The structural finding (MHC dominance) holds across both.

What this means

About 60% of the EUR-cohort schizophrenia GWS SNPs that 23andMe v5 types fall in the chr 6 MHC region — one extended LD block. Any unweighted chip-PRS computed against the EUR cohort is dominated by your tag-SNP dosage at this single block, not by polygenic accumulation across independent loci.

The same chip cross-referenced against the East Asian-cohort schizophrenia GWS at current sample sizes hits zero MHC SNPs — not because the MHC has no biological role in EAS schizophrenia, but because the specific tag SNPs that reach genome-wide significance in EUR populations have minor allele frequencies under 1% in East Asians (Lam et al. 2019: rs13194504 EAS MAF <1% vs 9% EUR; the C4-BS allele uncommon in Han Chinese). The 9-Mb MHC band is empty.

Implication for chip-PRS users: report MHC and non-MHC partitions separately, especially when computing on EUR-cohort GWAS. If your ancestry is neither EUR nor EAS (e.g., African, South Asian, Latino, or admixed), neither column above is the right reference for you — the takeaway here is that GWS-tag-SNP MHC dominance is ancestry-cohort-specific, not that one of these two cohorts applies to everyone. The companion methods paper develops this in detail.

Caveats