Roast Moonlight: a free site that lists the medicines linked to a disease, each with a source
retained text
We have built Moonlight: a free site that lists the medicines linked to a disease, each with a source Link: [https://www.moonlightclues.com](https://www.moonlightclues.com) What it is: you type a condition and get the medicines linked to it, sorted into approved, being studied, and "new clues" (medicines that act on proteins linked to the disease but haven't been tested in it). Each card has a source: an FDA label, an EMA page, a trial or a research record. It is a window onto one open database, Open Targets, so its gaps are their gaps, and "you missed X" is exactly the help we want. It's free and still evolving. We're a small team in Kochi, India. How it's built: Python on Cloud Run, pulling from the Open Targets GraphQL API, with 1,000+ tests. What actually hurt, all of it data quality rather than code: Indication wording. "Post-polycythemia vera myelofibrosis" was being read as polycythaemia vera. Salt forms. "Givosiran" and "givosiran sodium" show up as two medicines unless you merge them. Approvals attached to ontology parent terms, so a drug approved for a broad group looks approved for a narrow disease inside it. Dead DailyMed links. A protein filter we added to cut noise quietly dropped real trials. Our conclusion is that the moat isn't the data, which is open. It's the corrections layer: every fix a patient or doctor sends us. Traction, honestly: we added analytics on 16 Nov. Since then there have been 286 visitors and 444 page views, with about 81% bouncing. Referrals come mostly from Reddit, then Instagram. The US sends the most visitors, then India. Roast it, and I'd especially like builders' advice on: 1. How do you turn "you missed X" comments into a corrections pipeline people trust, without hand-checking every one? 2. Is an 81% bounce rate on a lookup tool bad, or just people getting their answer and leaving? 3. Who should this be for first: patients, carers or clinicians?
document fetched 2026-11-20 00:25 · canonical https://reddit.com/r/developersIndia/comments/1xzzbac/roast_moonlight_a_free_site_that_lists_the