Founder's note
About Varia
What I felt was missing from the consumer genomic space and what I built.
Hello,
I appreciate you taking a look at Varia. My goal is a genuinely useful tool, built on strict validation and fact-checking, that turns your raw genome into reliable, factual insight. No supplement suggestions, no trendy nonsense, only auditable, peer-reviewed information, so you can see clearly where something may be worth acting on and take it to your doctor with confidence.
I have had a semi-obsession with health and well-being for most of my life. For a long time my problem was that I did not have the right mentors and did not know where to look. I "Googled around" for medical and health guidance; some of you may be familiar with that tactic.
I ran my 23andMe genome without a specific reason, mostly because I thought it would be interesting to have it scanned and see if any surprises turned up. The results were interesting but surface level. It felt a little gimmicky unless you got real value from the ancestry and family-connection side. Years later I started listening to people like Peter Attia, Rhonda Patrick, and Andrew Huberman. Around 2018 or 2019 I learned that Rhonda Patrick offered a tool that would take your raw 23andMe data and return a detailed report covering parts of your genome 23andMe never showed you. That is where I found out I am APOE 4/4. That single discovery cemented my interest in longevity, and with it, genomic interpretation.
My problem was that, while my genome was not changing, the science was, and the tools available to me were not good enough to get the information I wanted in a form I could actually act on. I eventually had a 30x whole-genome sequence done through Sequencing.com and was excited to use their Genome Explorer. It helped, but I was surprised how often an rsID search came back with zero results. I knew the variant was in my data; I could not understand why the scanner would not find it. The mass-market consumer tests felt the same way: heavy on gamified traits and ancestry, while some of the variants that actually matter for a genotype like mine were either missing or buried.
So I downloaded my raw WGS data and worked out a way to reliably read and pull specific variants from it. I stopped caring why the other companies were not offering a comprehensive scanner; I knew I could build one. That scanner is now the engine inside Varia. Because its process is rigorous and fully transparent, you can see exactly how it reads your file and confirm that it accurately reports your genotype for every variant Varia covers.
Ultimately, I built Varia for the question I actually had: given my genotype, what should I be ready to discuss with my doctor at my next appointment? Varia's answer is a curated set of 64 variants across 6 health areas, kept current as the literature evolves, with the editorial discipline made visible so you can decide for yourself whether to trust it.
The 6 health areas are my own organizing choice, not an official medical standard. They are where I found the evidence strong enough, and actionable enough, to be worth curating: Heart & Lipids; Brain & Cognition, including the Alzheimer's-related biology my own APOE result sent me chasing; Metabolism & Longevity, which folds in blood sugar, methylation, inflammation, and telomere biology; Medication Response, where your genes shape how you process common medications; Cancer Risk, common low-penetrance cancer associations from published studies; and Autoimmune, common autoimmune associations with honest limits on what consumer files can assess. As the science grows, these areas will grow with it.
Rather than handing you an endless sea of distracting, mostly uninteresting data, Varia offers only vetted findings. My aim is a centralized, reliable, and growing resource for people who want to explore their own genome in genuinely actionable ways. I am not trying to convince you of anything. Every editorial choice on this site assumes you are making decisions that matter.
I am one person doing this work. If you believe something here is wrong, email me at [email protected]. I promise I will look into it and apply any correction that holds up.
Eric
Founder, Varia
Knowing both what a tool does and what it deliberately doesn't do is the trust signal. Here is what Varia is, and what it is not.
What Varia is
Editorial discipline made visible. Every insight in Varia has a named significance tier (ELEVATED, HIGH, MODERATE, PHARMACOGENOMIC, PROTECTIVE, NEUTRAL, UNRESOLVED), a named source from an allowlist of peer-reviewed journals and authoritative bodies, and a documented review process. When the literature is mixed, the entry shows an "Evidence is mixed" callout instead of silently picking a side. When a citation is in process or recently resolved, the entry shows a provisional flag rather than displaying a stale interpretation. The full standard is published at Editorial Standards.
A scanner that tells you what it saw. Most consumer tools skip variants their input did not test and let you assume silence means good news. Varia labels every finding as "tested, you carry it," "tested, you do not carry it," or "not tested in your file" so you always know what you actually learned. The full reading process is under the Scan Types and How Varia Reads tab.
Pharmacogenetics as the focal use case. Knowing your CYP2C19 status before clopidogrel, your SLCO1B1 status before a statin, or your VKORC1 status before warfarin can change the clinical decision in front of you. Varia surfaces every pharmacogenomic-tier finding into one cross-domain panel called Medication Response, and produces a clinician-readable PDF ordered PGx-first so you can carry it into a prescriber appointment. The limitations are named, not hidden: consumer-chip data identifies SNP variants in PGx genes but does not detect CYP2D6 copy-number variation or full HLA typing. See Medication Response.
Transparency about authority. Every source Varia cites is publicly listed at Sources: peer-reviewed journal allowlist, variant catalogs, clinical guidelines, domain consortia, build-time tooling. The page is a reference list, not a curation dump. Where a recognized authority is missing, the reason is named.
What Varia is not
Varia is not an AI chat assistant for your genome. Conversational genomics interpretation is the failure mode of every consumer product that has tried it. It generates infinite exploration without curation. Varia gives you a fixed, peer-reviewed knowledge base instead.
Varia is not a polygenic risk score engine. PRS validity is population-stratified in ways that consumer-genomics products handle badly, and the most-defensible PRS panels are clinical-grade products in regulated markets. The risk of being wrong outweighs the benefit of being timely.
Varia is not a supplement, lifestyle, or training protocol generator. Making product or behavior recommendations crosses a deliberate line. The Varia Genomic Brief is the bright edge of what Varia publishes: variants and biomarker and drug-class names, never doses or protocols. Those decisions belong to you and your doctor.
Varia is not a continuous-tracking or wearable-integration product. Continuous engagement is a different product shape and a different commercial model. Varia is a one-time interpretation tool, kept current by database updates rather than by your data stream.
Varia is not a whole-genome exploration tool. Exploration is the explicit anti-positioning. Varia covers what is currently understood and clinically actionable. It does not help you wade into what is not.
Varia is not an ancestry product. Ancestry is a different category (23andMe, AncestryDNA, LivingDNA) with different data requirements and a different competitive set.
Two questions come up a lot: what kind of DNA file works, and how Varia actually reads it. Here is the short version of both.
The file formats Varia reads
Not every DNA test reads the same slice of your genome, and the difference changes what Varia can tell you.
SNP arrays are the consumer chips behind 23andMe and AncestryDNA. They read roughly 600,000 to 900,000 pre-chosen positions, a fraction of a percent of the genome. If a variant matters but wasn't on the chip, the chip can't see it.
Whole exome sequencing reads the protein-coding regions, about 1 to 2 percent of the genome, where much of the disease-relevant variation has been characterized. It misses the non-coding regions where the science is still maturing.
Whole genome sequencing at 30x reads all 3 billion bases. It's the most complete consumer-accessible format, delivered as a VCF file, and it's the one Varia is built around. Varia accepts all three.
Coverage isn't the same as quality. A bigger file gives Varia more to work with, but it doesn't change what the literature supports. The curation bar is the same either way; a higher-coverage file just leaves fewer positions marked "not tested in your file."
How Varia reads it
A raw file isn't a finished interpretation. The same data, read with different assumptions, can produce different answers, so Varia's reading is built around the cases consumer genetic data trips over most.
It builds compound results correctly. Some findings depend on more than one variant together, like your APOE type, which only means something as the pair, not the two positions read separately. Varia composes the pair and flags it when only one half is present.
It tells "you don't carry it" apart from "we couldn't check." When a clinical lab returns a variants-only file, a missing position almost always means you match the reference, not that the position went untested. Varia reads that correctly instead of turning a gap into false reassurance, the failure consumer genetics is most criticized for.
It handles files as they actually arrive. Chip exports and clinical VCFs come in different genome builds and conventions; Varia detects the build, matches by reference ID, normalizes the strand, and refuses a file it can't read rather than guessing. What Varia chooses to surface is at Editorial Standards; how medication-response findings surface is on Medication Response.
If you want to go deeper
Pick the depth page that matches the question you are bringing.
- Methodology. The tier system, source allowlist, and recommendation boundary that drive every finding.
- Editorial Standards. The full editorial discipline, including how I handle provisional citations and conflict-of-evidence.
- The Genome. What a genome is, what scan types cover, and how I built the scanner.
- Medication Response. The pharmacogenetics focal use case in clinical detail.
- Sources. The peer-reviewed journals, variant catalogs, clinical guidelines, and consortia I cite.
- Varia vs Promethease. A direct comparison with the dominant raw-data interpretation tool.
Varia is one person's editorial mission, priced fairly, with the work made visible. If something here speaks to a question you have been carrying, the scan is the next step.
