The GenoSphereTM connector lets Claude answer population-level questions about human genetics and health directly from the Helix GenoSphere research dataset: a de-identified cohort of more than 500,000 Exome-sequenced participants with linked electronic health record data. Ask how many people carry pathogenic variants in a gene, how common a condition is in the cohort, or what the dataset covers, and Claude queries Helix and answers with aggregate statistics.
It is built for researchers, clinical scientists, and life-sciences teams who want to explore the dataset conversationally without writing SQL or learning an internal schema.
What you can ask
The connector is strongest at population and prevalence questions across major therapeutic areas including cardiovascular, metabolic, neurology, and many others. Some example queries:
- "How many participants carry pathogenic BRCA1 variants?"
- "Is the LDLR gene covered in this dataset?"
- "How many people in the cohort have a type 2 diabetes diagnosis?"
- "Find the concept for hypertrophic cardiomyopathy and tell me how common it is."
- "Give me an overview of the whole dataset: size, ancestry groups, and what clinical data is linked."
Claude will pick the right tool, chain them when needed (for example, resolving a condition name to an internal concept before counting patients), and report what the data shows.
What it returns, and what it does not
Every result is a de-identified aggregate statistic: a count or a percentage over the cohort. The connector never returns individual records, genotypes tied to a person, or any protected health information. It is read-only, and small counts are suppressed to protect privacy.
Getting access
Access is granted to named, agreement-bound parties rather than the general public.
- Request access by contacting lifesciences@helix.com. Helix provisions your account in its identity provider.
- In Claude, open the connector directory and enable Helix GenoSphere.
- The first time you use it, you are redirected to sign in through Helix's Okta login. Approve the access request to complete the connection.
Once connected, you can begin asking questions in any Claude conversation where the connector is enabled.
If you enable the connector without a provisioned account, you will see a message directing you to contact Helix rather than any data.
Troubleshooting
The connection fails or you are not redirected to sign in. Make sure you completed the Helix Okta sign-in and approved the access request. Disabling and re-enabling the connector restarts the sign-in flow.
You connect but every query returns an access or permission error. Your account may not be provisioned yet, or your access may have lapsed. Contact lifesciences@helix.com to confirm your account is active.
A query returns a suppressed, zero, or empty result. This is expected, not an error. Small groups are masked to protect privacy, and a zero or empty result means nothing matched within the dataset's scope rather than that the answer is definitively none. Try rephrasing, or ask Claude to search related terms or domains.
A gene comes back as not available. The queryable panel is a curated subset. Helix holds exome-wide data beyond it, so a gene's absence from the panel does not mean Helix lacks data on it. Contact lifesciences@helix.com about coverage for a specific gene.
Results look slow or time out. Most queries return in a few seconds. If a request times out, retry once; if it persists, contact support.
Support and licensing
For access requests, account issues, dataset coverage questions, or licensing of the full GenoSphere dataset, contact lifesciences@helix.com.
Learn more about Helix GenoSphere at helix.com/life-sciences/genosphere.