How it works,
and why it's private.
Open research, run inside your browser. Your genome doesn't need to travel to someone else's server to tell you something about yourself.
Your DNA never
leaves your browser.
Your DNA is the most private and personal data you have. It's uniquely yours, it never changes, and it reveals deeply sensitive information about your health, your ancestry, and your family. Once it's compromised, there's no resetting it like a password.
The DNA testing industry has seen major data breaches affecting millions of people. When companies store your genetic data on their servers, it becomes a target — and once that data is exposed, the damage is permanent.
Every calculation on this site happens locally in your browser. Your DNA file is never uploaded, never transmitted, and never stored on any server. We can't see it, we don't want to see it, and our system is designed so that we never could.
Your DNA stays on your device. That's not a feature — it's the whole point.
The method,
in plain terms.
What is a polygenic score?
A polygenic score summarises the combined influence of many genetic variants across your genome on a particular trait. Each variant has a small effect, but together they indicate where your genetic profile sits compared to a reference population.
What is the PGC?
The Psychiatric Genomics Consortium (PGC) is the largest international consortium conducting genome-wide association studies for psychiatric conditions. Their summary statistics form the basis for the scoring files used by this tool.
Why isn't this a diagnosis?
A polygenic score typically accounts for 1–10% of the overall picture. The rest comes from environment, life experiences, and other complex factors. A higher score doesn't mean you have a condition, and a lower score doesn't mean you don't.
How are scores calculated?
Your polygenic risk score is the sum of dosage × beta across all matched SNPs, where dosage is the number of effect allele copies (0, 1, or 2) and beta is the log odds ratio from the GWAS. To make the raw score interpretable, we convert it to a population percentile using a z-score: expected contribution per SNP is 2 × freq × beta and variance is 2 × freq × (1 − freq) × beta². These are summed only over SNPs that matched your file, so coverage is automatically accounted for.
How do we validate the scores?
We validate our normalisation formula against a reference panel of 503 European individuals from the 1000 Genomes Project (Phase 3). For genome-wide significant SNPs — which are approximately independent across the genome — the theoretical formula matches the empirical distribution to within 0.3%. For detailed reports with millions of SNPs at relaxed significance thresholds, LD can inflate the true variance; the empirical reference panel is essential for accurate percentiles there.
What is LD clumping?
Nearby genetic variants tend to be inherited together. Without LD clumping, correlated variants can be counted multiple times, slightly inflating scores. This tool does not perform LD clumping — scores should be interpreted with this in mind.
Sources &
acknowledgements.
This tool would not be possible without the generous open sharing of research data.
Psychiatric Genomics Consortium (PGC)
GWAS summary statistics for ADHD, Depression, Bipolar Disorder, OCD, Autism, and Anxiety. The PGC is the largest international consortium studying the genetics of psychiatric conditions.
pgc.unc.edu →OpenMed Genomics
PGC data was accessed via the OpenMed project on HuggingFace, which makes genomic research data more accessible. Licensed under CC BY 4.0.
huggingface.co/OpenMed →GWAS Catalog (EMBL-EBI)
Cross-trait association data used in detailed reports. A curated collection of all published genome-wide association studies. Public domain (CC0).
ebi.ac.uk/gwas →ClinVar (NCBI)
Clinical significance annotations for genetic variants used in detailed reports. A freely accessible archive of reports on the relationships among human variations and phenotypes. Public domain.
ncbi.nlm.nih.gov/clinvar →Gene Ontology Consortium
Biological pathway and gene function annotations. A computational representation of our knowledge of how genes encode biological functions. Licensed under CC BY 4.0.
geneontology.org →GENCODE
Gene coordinates and annotations used to map genomic regions to genes. Reference gene annotation for the human genome. Licensed under CC BY.
gencodegenes.org →NCBI Gene
Gene descriptions and summaries used alongside GENCODE coordinates. Public domain.
ncbi.nlm.nih.gov/gene →Get in touch.
Have a question, a data source we should know about, or feedback on the methodology? We'd love to hear from you.
questions@privatedna.health →Colophon.
Made by Pete Nelson (petenelson.co.uk) with Claude, Cloudflare and Coffee.