codegen.eu · raw DNA analysis

Genetic research engine

A deep-research report, tens of thousands of pages long, on your own DNA - your genome analyzed against 4,000+ reports (diseases and traits) and the scientific literature, using frontier models that processed over 1.5 trillion reasoning tokens across recent scientific papers. Search it like documentation. Summarized, ranked for relevance and grouped into reports with polygenic risk scores.

Start FREE See a sample report

Free to start: search every variant and read its short summary. The full report is €24 + VAT, one payment, no subscription.

Zero data retention

Coming back? There is no password. Upload the same raw data file again: your report is rebuilt, your favorites come back and anything you paid for stays unlocked. How logging in works

coronary artery disease

Your score vs population

lower riskhigher risk

rs10757278 W:2.40 49%

9p21 risk locus (G/G). A common variant that many studies associate with coronary artery disease.

What you get

Raw DNA analysis you can actually use.

Upload the raw data file you already have. Files from 23andMe, AncestryDNA, MyHeritage, FamilyTreeDNA and others are supported; whole-genome sequencing files are not supported yet.
One searchable report. Your genotypes are matched against a catalog of 100k+ curated SNPs, variant groups and 4,000+ sub-reports, one per disease or measurement; every finding is scored against the general population, summarized in plain language and linked to the papers behind it. Finding your way around codegen.eu
Start FREE by reading all the short summaries. A single payment of €24 + VAT, with no subscription, unlocks the rest for good, with unlimited updates.
Your raw file is never stored. We process it in memory and delete it within seconds; your report data is deleted when you log out or after 6 hours idle.

Reports

What studies report, from cancers to caffeine.

Common diseases. Where your polygenic score sits, compared with the tested population, for type 2 diabetes, coronary artery disease, high blood pressure, atrial fibrillation and stroke; breast, prostate and colorectal cancer and melanoma; Alzheimer's and Parkinson's; celiac disease, rheumatoid arthritis, psoriasis, asthma and osteoporosis. A research-grade signal, not a diagnosis. Find out more: Modern diseases: cancers, heart disease and metabolic syndrome
Metabolism and nutrition. Variants that studies associate with LDL cholesterol, vitamin D and B12 levels, caffeine and lactose, body mass index, body fat, lean body mass and base metabolic rate. Only a lab test measures your actual level. Nutrigenomics: lactose, caffeine, vitamins and your genes
Genes and medicines. What research reports about genes and common medicines, such as warfarin and statins. Never change a medication without talking to your prescriber.
Psychological and everyday traits. Variants that studies associate with depression, anxiety, insomnia and sleep duration, migraine, resting heart rate, grip strength, physical activity and energy intake, male-pattern baldness.
Genes people ask about. APOE, MTHFR, HFE, Factor V Leiden (F5), LCT, CYP1A2, FTO, COMT and ACTN3, searchable by gene name or rs-ID.
Breadth. 4,000+ reports across 15 categories: common diseases, metabolism and nutrition, genes and medicines, psychological and physical traits.

Polygenic risk scores

Scores computed by the book.

A polygenic risk score per established disease or measurement. GWAS associations are harmonized onto one effect allele, meta-analyzed across studies by inverse-variance weighting and reduced to a per-allele weight.
Your standing. Your allele dosages are summed, standardized against the population and ranked as an empirical percentile. Understanding your reports
Honest about thin evidence. Reports resting on few variants are shrunk toward average and marked low confidence.
Published method. Grounded in science: follows the peer-reviewed protocol of Choi, Mak and O'Reilly, Nat Protoc 2020. A relative signal, not a diagnosis.

Geno Researcher

Ask about your report in your own words.

An AI assistant for questions about your own genetic report. It looks through your report for the variants that matter to the question, answers from them and lists the ones it used, each linked to its summary, so you can check every point.
Attach a lab test. Add a lab test result or a doctor's letter as a PDF and look for connections between the two. We remove personal details from the text before anything leaves our server.
Only what the question needs. The AI gets your question and the matching entries of your report, never your genome hash or raw DNA file. We send it only to AI providers bound by zero data retention terms, so they don't keep the request data. Conversations are erased when you log out.
Included with the full report. Unlocking all your reports includes credits for your questions.
Every answer shows its sources. Answers are written by AI and can contain mistakes, so check anything important against the variants they list. It explains and connects what your report already contains and doesn't diagnose or prescribe. How to use it, with a real example

Search it like documentation.

Type a disease, drug, gene or rs-ID. Results come back ranked for relevance.
Narrow them down. By category, by genotype frequency (common, infrequent, rare, very rare) and by risk band (good, info, warning, bad). Searching the full report
Keep what matters. A rare-variants page collects your low-frequency genotypes; star anything to keep it in your favorites.

Summaries

Summaries from current research.

Practical takeaways. A plain-language summary for genotypes, assembled from recent peer-reviewed papers plus dbSNP, the GWAS Catalog, ClinVar and OMIM. Starts with simple explanations and expands in depth, facilitating research for curious users.
With the numbers that matter. An impact score (0 to 4+), your genotype's frequency in the population, the gene and strand, and the references behind it. Variant geno card explained
Built from the original research papers, not second-hand summaries. Research current to 2026, from a deep-research pass over a vast corpus of papers exceeding 1.5 trillion tokens of reasoning and input.

Variant groups

Some variants only make sense together.

Capture complex interactions. A haplotype, or several variants in the same gene whose combined pattern is what matters.
Read like a single genotype. A variant group has its own impact score and references and shows up in search and report results next to ordinary genotypes.