“When you’ve spent years trying to understand a health issue, the question you bring to your DNA is personal. Could anything in this file help me make sense of what I’ve been experiencing? That’s the question we built Mutant to help people explore.” - Mutant Genomics team
Millions of people hold a raw DNA file from a consumer test and never learn much from it. The reports that come with such files often explain one variant or one trait at a time, which leaves anyone with persistent symptoms or a condition that runs in the family without a clear way to connect those results to their own health picture. Mutant Genomics, a product of Mutant Biotech, is built around that gap, and its latest step is a plugin that carries its analysis into ChatGPT Health.
The company announced the ChatGPT plugin on 11 November 2026 from Austin, Texas. The platform lives at mutantgenomics.com, and the plugin is available in the ChatGPT plugin directory. Access to ChatGPT Health itself depends on OpenAI's availability and account requirements.
Weighing each hypothesis inside ChatGPT Health #
The plugin is where the platform meets everyday behaviour. For users with access to ChatGPT Health, it brings genetic hypotheses into a conversation where broader health information is available. A person can weigh each hypothesis against the symptoms, lab results and history they choose to connect.
Mutant Genomics offers an example of the kind of question this supports: “Which Mutant findings fit my symptoms and lab results? What evidence points away from them, and what should I discuss with my clinician?” The framing matters. A hypothesis may fit some observations and conflict with others, and the available information may be insufficient to assess a connection at all. Exploring those uncertainties can help someone arrive at a clinical follow-up with more focused questions.
A YC startup is betting $1 million that AI can crack five materials science puzzles Discovered Materials, an IIT Madras-founded YC startup that raised $9 million in August, is offering $200,000 per problem for breakthroughs on cuprate superconductivity, silicon interfaces, and three other physics puzzles. - AI startup materials science million dollar challenge - breakthrough materials discovery AI competition prize money
The platform also covers cases where the DNA file falls short. When a file lacks the markers needed to assess a hypothesis, people with access to ChatGPT Health can still explore the underlying health question using the symptoms, lab results and medical history they choose to share. Users can also work in regular ChatGPT conversations using health details they provide, or review the analysis directly in the Mutant portal.
A family angle is built in as well. With a relative's consent, a user can discuss the family history that person chooses to share alongside their own Mutant findings. The relative's connected records stay in their own ChatGPT Health account, and Mutant does not receive or store them.
From a DNA file to ranked hypotheses #
The idea is simple to state. Someone brings DNA data they already have, and Mutant's Genomic Hypotheses Engine turns it into a ranked list of educational hypotheses about possible connections. For each one, the person can see why it was raised, which variants contributed, what research supports it, and where the file leaves gaps.
No new test is needed. The platform accepts raw DNA text files from 23andMe and AncestryDNA, as well as supported whole-genome VCF files, including compressed ones. Microarray files cover selected genetic markers, so some hypotheses may have limited genetic evidence or be impossible to assess from the file. Whole-genome sequencing files offer the broadest coverage, and a user can start with the data they have and add more complete genome data later.
Before adding a hypothesis to its catalogue, Mutant uses an AI-assisted research workflow to find relevant studies, identify the variants and biological pathways involved, and organise the evidence for review. The review distinguishes between an experiment that shows how a variant affects a biological process, a study that finds a statistical association, and a computational prediction. Once a hypothesis is in the catalogue, the engine evaluates it using the variants covered by each person’s DNA file. When several findings point toward the same biological process, the report shows how they fit together.
Evidence, and the limits of evidence #
The ranking reflects the genetic evidence available in the person’s file and the strength of the research behind it. Population frequency helps put a variant’s rarity in context, and Evo 2 helps refine the weight of selected variants. Where a relevant ClinGen curation exists, Mutant includes it as additional clinical context alongside the score. ClinGen’s curations are separate from the ranking calculation.
What stands out is how much space the platform gives to what it cannot say. Missing data is not treated as a normal or protective result, and coverage is shown alongside each analysis. A higher rank means greater priority for exploration within a person's own analysis. It is not the probability that a condition is present. Mutant states plainly that genetic susceptibility is not certainty and that association is not causation.
OpenAI's AI pipeline turned 2,308 Linux kernel bugs into 948 root exploits OpenAI's cyber research pipeline reproduced 1,724 real Linux kernel bugs from 2,308 candidates and turned 948 into validated root exploits with patches, for roughly $1.2 million in a single run. - openai ai pipeline linux kernel root exploits - how openai found 948 linux kernel exploits
The company is equally clear about its role. It provides educational and informational analysis, and its findings do not establish diagnoses or causes of symptoms. Medical decisions, it says, require appropriate professional evaluation and clinical confirmation. Selected inherited findings, including PRSS1-related pancreatitis risk, need clinical confirmation and context.
Control over data #
Health and genetic data raise obvious privacy questions, and the company addresses them in its documentation. Mutant processes the original DNA file in the user’s browser. The selected genetic data needed for the analysis is sent to Mutant and stored with the resulting findings, and Mutant states that it does not sell genetic data.
The ChatGPT connection is opt-in. Users review and approve the genetic-data processing and result-sharing permissions during setup, and they can disconnect from the ChatGPT permissions screen. Mutant does not automatically receive a user's full ChatGPT conversation or connected health records. After signing in, account controls let people clear a report's DNA data and derived analysis, or request account closure to delete their Mutant-stored data. Deletion does not remove findings already shared in a ChatGPT conversation or another service.
The company also warns users about the tools around its product. AI can help organise information and generate questions, it says, but it can make mistakes, so Mutant findings and AI-generated interpretations should be reviewed critically and should not replace professional medical care.
Who it is for, and what it costs #
Mutant describes its audience as people who want more from DNA data they already have, people exploring complex health questions where symptoms, lab results and history do not fit a simple explanation, and people who use AI to organise their health information. Practitioners and researchers who want a transparent view of the evidence, coverage and limitations behind a result are named as well. No genetics background is required.
Access is split into two tiers. Mutant Free costs nothing, needs no credit card, and includes a person's top three ranked health hypotheses in full, with evidence breakdowns and related findings, usable in the portal or through the ChatGPT plugin. Mutant Full unlocks every ranked hypothesis, full evidence breakdowns, contributing patterns and supporting variants for a one-time $49 for founding members, with ongoing Full access and no recurring charges. Visitors can also explore an interactive sample analysis before providing any DNA data of their own.
Why the approach matters #
Direct-to-consumer genetics has long struggled with a communication problem. A list of variants can feel authoritative while explaining very little, and a single trait report can invite conclusions the evidence does not support. Mutant's answer is to organise findings, rank them for exploration, and keep the evidence and its gaps in view, then hand the user a way to test each idea against their own context.
Whether that approach changes how people prepare for conversations with clinicians will depend on how they use it. The design choices point in a careful direction: educational language rather than diagnostic claims, visible coverage, opt-in connections, and data the user can remove. For anyone sitting on a DNA file and a set of unanswered health questions, the platform offers a structured place to start, and the sample analysis is open to browse without an account.
Join the discussion #
Open in the community → Almost there. Sign in and your reply posts straight away.