How Gosset works
From the world’s sources to answers you can cite
Gosset reads clinical trial registries, regulators, publications, patents, deals and news every day, links them into one knowledge graph, and structures it with ontologies of targets, modalities and diseases. That is what lets it answer research questions with a source behind every line.
Updated . Every type of source refreshes daily; this page updates hourly.
1 Sources
Six types of source, read every day
51 sources across 17 countries, 7.2M records, and about 5,351 new records a day.
Clinical trial registries
Up to date17 registries across 14 countries
947,846 records
about 446 added per day
Live · newest record
Regulators
Up to date14 regulators across 11 countries
577,905 records
about 915 added per day
Within a day · newest record
Publications
Up to date5 publication indexes
1,979,090 records
about 2,951 added per day
Live · newest record
Patents
Up to date8 patent offices across 6 countries
1,051,720 records
about 512 added per day
Within a week · newest record
Company materials
Up to date28,768 company websites read for pipelines and investor materials
1,629,777 records
about 382 added per day
Within a day · newest record
News
Up to date1 news feed
1,030,957 records
about 145 added per day
Within a day · newest record
2 Knowledge graph
One graph of drugs, trials, companies and deals
Every record is resolved to the drug, trial, company or deal it is about, duplicates across sources are merged, and the results are connected: which trials test a drug, who develops it, and which deals changed its hands.
- Drugs
- 155,738
- Drug intervention trials
- 464,258
- Companies
- 193,004
- Deals
- 324,856
Drug intervention trials, deduplicated across registries: only trials that test a drug or biologic are kept, so there are fewer than the registry records above.
3 Ontologies
Targets, modalities and diseases hold it together
Three ontologies give every drug and trial a common language, so a question about a target or a disease finds every program however a source spelled it. A concept is assigned only when a source document supports it, so coverage counts evidence, not guesses.
Targets
8,795 targets
79% of drugs have a source-backed target
Genes and proteins, e.g. TL1A, B7-H3, Claudin 18.2
Modalities
557 modalities
67% of drugs have a source-backed modality
How a drug works, e.g. ADC, bispecific, small molecule
Diseases
7,163 diseases
95% of trials have a source-backed disease
Indications, e.g. small cell lung cancer, ulcerative colitis
4 Answers
Research questions, answered with sources
Ask in plain language. Gosset queries the graph and writes a report in which every claim links to the record it came from.
- “What’s the competitive landscape of TL1A?”
- “Which B7-H3 ADCs have Phase 3 trials in small cell lung cancer?”
- “Which China-origin ADCs now in Phase 2 or 3 have licensing agreements with US partners since 2023?”
From the report China-to-US ADC licensing BL-B01D1 (izalontamab brengitecan) went from SystImmune to Bristol Myers Squibb in a global collaboration announced on 11 December 2023[1], and is now in Phase 3 in China for triple-negative breast cancer, oesophageal squamous cell carcinoma and nasopharyngeal carcinoma[2]. Hengrui licensed its DLL3 ADC SHR-4849 to IDEAYA Biosciences at Phase 1 on 29 December 2024[3].
- SystImmune and Bristol Myers Squibb announce a global strategic collaboration for BL-B01D1 investors.bms.com
- Izalontamab brengitecan improves overall and progression-free survival in TNBC and ESCC news.bms.com
- IDEAYA enters exclusive license with Hengrui Pharma for SHR-4849, a Phase 1 DLL3 ADC media.ideayabio.com
Times are relative to now; hover a time for the exact UTC timestamp. “Newest record” is the most recent record we hold of that type. Per-day figures are averages over the last 30 days.