Skip to main content
DNA Romance
DNA Romance
Sign Up

DRom 1.0 and DRom 2.0: How DNA Romance Predicts Chemistry and Phenotype

Dr. Timothy SextonDNA Romance
  • genetics
  • mhc
  • dna-romance
DRom 1.0 and DRom 2.0: How DNA Romance Predicts Chemistry and Phenotype

DNA Romance runs two production algorithms over the raw DNA files our users upload. DRom 1.0 predicts romantic chemistry — the gut-feeling attraction that determines whether two people who have only chatted online will click in person. DRom 2.0 predicts phenotype — the observable physical and biochemical traits that describe what a person is actually like. Together they turn a 23andMe or AncestryDNA file into something useful for matchmaking.

This is a product note. If you want the underlying biology, our companion articles cover the love genes, HLA typing and the MHC, and the body-odour gene in detail. Here we describe what each algorithm does, what evidence supports it, and what it does not pretend to do.

DRom 1.0 — Predicting Romantic Chemistry

What "chemistry" actually means

In conventional dating, "chemistry" is the warm, slightly inexplicable feeling that hits when two compatible people first meet face-to-face. It is part smell, part body language, part voice, part subconscious threat assessment, part something we still cannot fully name. You either feel it or you don't.

That is fine when you meet someone in person on day one. It is a problem in online dating, where two people might exchange messages for weeks, build up an idea of each other, fly across the country to meet — and then discover, in thirty seconds, that there is nothing there. All that effort, gone. This is the central frustration users keep telling us about: too much guesswork, not enough signal.

DRom 1.0 exists to put real signal into the front of that funnel.

How DRom 1.0 works

DRom 1.0 reads 100 Single Nucleotide Polymorphism (SNP) markers positioned in candidate genes inside and around the Major Histocompatibility Complex (MHC) — the love genes. These SNPs were selected because they tag the kinds of HLA-allele diversity that the foundational Wedekind et al. (1995) sweaty-t-shirt study and follow-up work suggest matter for partner choice through scent.

The algorithm computes how different two people's profiles are across those 100 markers. The principle, drawn from decades of work on MHC and mate choice, is straightforward:

People with more dissimilar immune-system genes find each other's scent more pleasant, and tend to form more lasting relationships.

DRom 1.0 returns a single compatibility score per pair. Higher scores correspond to greater HLA-region dissimilarity — a candidate signal for chemistry. We have written separately about why HLA diversity matters and where the underlying scent-and-mate-choice evidence is strong versus where it is more contested. The honest summary is that the MHC effect on attraction is real but not the whole story, and DRom 1.0 is one input among many — not a verdict.

What DRom 1.0 is good for

  • Pre-screening. Before you spend weeks chatting with someone, you have a rough biological prior on whether the in-person meeting is likely to feel right.
  • Reducing decision fatigue. With a score in front of you, you can prioritise messages, dates, and follow-ups with more confidence.
  • Talking to your matches like adults. "Our DRom 1.0 score was high, want to grab a coffee?" is a more honest conversation starter than most pickup lines.

The chemistry score lays a foundation. The rest of compatibility — values, hobbies, life stage, sense of humour — is on you and your match to discover.

DRom 2.0 — Predicting Phenotype from DNA

What is a phenotype?

A phenotype is any observable trait an organism actually displays — eye colour, hair colour, height, blood type, lactose tolerance, pet allergy susceptibility, propensity for caffeine sensitivity, and many others. Phenotype is what you can see; the underlying DNA — your genotype — is the instruction manual that produced it.

Many phenotypes are heritable enough that they can be predicted accurately from DNA alone, especially when the underlying biology is well understood (eye and hair colour, for instance, are governed by a relatively small number of well-characterised SNPs). Other traits — height, body shape, complex personality features — depend on hundreds or thousands of variants plus environment, and are much harder.

How DRom 2.0 works

DRom 2.0 is a trained neural network that reads SNPs across the genome and outputs predicted values for a curated set of phenotypic traits. We use it for two complementary purposes:

  1. Autocompleting your dating profile. The fields that take new users the longest to fill in — eye colour, hair colour, blood type, allergies, lactose status — are exactly the fields DNA can predict. Users who upload DNA see a profile that pre-populates with high-confidence phenotype guesses, which they then confirm or override.
  2. Identity and consistency checks. A profile photo can be doctored or borrowed. A DNA file is much harder to fake. When DRom 2.0 predicts brown eyes from genotype and the photo shows brown eyes, that is a small but real consistency signal. When the prediction and the photo disagree dramatically, it is a flag worth investigating — though minor mismatches are often just incomplete predictions, not deception. We treat this as one input to our broader profile verification system, not a polygraph.

What DRom 2.0 will and won't do

DRom 2.0 is conservative by design. We expand the trait list when the underlying genetics is solid, the prediction is testable, and the trait actually matters for online dating. Future releases will widen the panel — coffee versus tea preferences, sensitivity to thrill-seeking sports, and a small number of additional phenotypes that real people care about when picking a partner.

What DRom 2.0 will not do:

  • Predict things genetics cannot reliably predict (personality types from DNA, for example — see our work on the Big Five vs MBTI for what personality testing can and cannot do).
  • Diagnose disease. Prediction of clinical conditions is the domain of regulated medical genetics, not a dating product.
  • Replace a real conversation. A DRom 2.0 mismatch between predicted hair colour and a recent profile photo could be deception — or could be that the user dyed their hair last month. Use the signal, do not weaponise it.

Using DRom 1.0 and DRom 2.0 Together

DRom 1.0 tells you whether a future first date is likely to spark. DRom 2.0 tells you what your match looks and lives like in measurable ways. Used together they shorten the gap between "matched online" and "this feels right in person" — which is, ultimately, the whole point of building DNA-aware matchmaking.

If you already have a raw DNA file from 23andMe, AncestryDNA, MyHeritage, or another consumer test, you can upload it to DNA Romance and have both algorithms run over it.

Explore your DNA-based compatibility → · Take a free personality test →


Related articles