API Workflows

Real-world tasks you can accomplish with the Amass API, each with a ready-to-run example. Use the outline on the right to jump to a specific scenario.

TrialCore

Map Clinical Landscape

How crowded is the GLP-1 space, and who's running what?

Search TrialCore for all trials targeting a condition, filtered by phase, status, and sponsor type. In one query you get back structured data on every active program: who's sponsoring it, what phase it's in, how many patients they're enrolling, and what their study design looks like.

curl "https://api.amass.tech/api/v1/cores/trialcore/records\
?query=GLP-1+receptor+agonist+obesity\
&phase=PHASE3\
&overallStatus=RECRUITING\
&interventionType=DRUG\
&limit=100" \
  -H "Authorization: Bearer amass_YOUR_KEY"

What you get: sponsorName, interventionNames, enrollment, designAllocation, designMasking, armGroups, startDate, facilityCountries — everything you need to build a competitive landscape table.

Widen the net across phases and statuses. The enum filters accept multiple values — repeat the param to match any of them (OR), while different filters still combine with AND. Capture both Phase 2 and Phase 3 programs that are either still recruiting or already fully enrolled:

curl "https://api.amass.tech/api/v1/cores/trialcore/records\
?query=GLP-1+receptor+agonist+obesity\
&phase=PHASE2\
&phase=PHASE3\
&overallStatus=RECRUITING\
&overallStatus=ACTIVE_NOT_RECRUITING\
&interventionType=DRUG\
&limit=100" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Example response (25 recruiting Phase 3 drug trials):

Track Competitor Pipeline

Who has posted results for their Phase 3 NASH trial?

Search by sponsor name and filter for hasResults: true to catch when competitors publish trial results. Pull the full outcome data with include=outcomes to see their primary and secondary endpoints, statistical parameters, and per-arm measurements — before the paper is even published.

curl "https://api.amass.tech/api/v1/cores/trialcore/records\
?query=NASH+nonalcoholic+steatohepatitis\
&hasResults=true\
&include=outcomes" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Example response (50 NASH trials with posted results — all with structured outcome data):

Drill into a single trial's primary outcome to see the actual per-arm measurements:

Elobixibat (Albireo, Phase 2, N=47) — Change From Baseline in Serum LDL (Least Squares Mean, mmol/L, Week 16)

Arm	Value	Spread
Elobixibat	-0.57	0.125
Placebo	-0.17	0.134

Learn From Stopped Trials

Why have trials in this indication been terminated?

Filter TrialCore for terminated or suspended trials. The whyStopped field captures the reason — safety, futility, enrollment challenges — giving you insight into pitfalls before you design your own program.

curl "https://api.amass.tech/api/v1/cores/trialcore/records\
?query=Alzheimer+disease+amyloid\
&overallStatus=TERMINATED\
&limit=100" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Example response (25 terminated trials, 22 with reason documented):

Benchmark Protocol

How have other Phase 2 trials in atopic dermatitis been designed?

Before writing your protocol, search TrialCore for completed trials in your indication. Analyze their study designs to establish benchmarks for your own trial.

curl "https://api.amass.tech/api/v1/cores/trialcore/records\
?query=atopic+dermatitis\
&phase=PHASE2\
&overallStatus=COMPLETED\
&studyType=INTERVENTIONAL\
&hasResults=true\
&include=outcomes\
&limit=100" \
  -H "Authorization: Bearer amass_YOUR_KEY"

What to extract: designAllocation (randomized vs. not), designMasking (blinding level), designInterventionModel (parallel, crossover), enrollment (sample sizes), primaryOutcomeMeasures, armGroups, and posted outcomes with actual measurements.

Example response (25 completed Phase 2 interventional trials with results — enrollment range 18–421, median 71):

Track Trial Results

Which immunotherapy trials in NSCLC have posted results?

curl "https://api.amass.tech/api/v1/cores/trialcore/records\
?query=non-small+cell+lung+cancer+immunotherapy\
&hasResults=true\
&include=outcomes" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Example response (25 trials with posted results):

Build Meta Analysis Datasets

Extract all outcome measurements for PD-1 inhibitors in melanoma.

Search TrialCore with hasResults: true and include=outcomes to pull structured, per-arm outcome data across dozens of trials. Each outcome includes paramType (mean, median, count), dispersionType (standard deviation, confidence interval), units, timeFrame, and individual measurements per arm — ready for quantitative synthesis without manual data extraction.

curl "https://api.amass.tech/api/v1/cores/trialcore/records\
?query=PD-1+inhibitor+melanoma\
&hasResults=true\
&include=outcomes\
&limit=100" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Spot Repurposing Signals

Is metformin showing up in trials outside of diabetes?

Search TrialCore for a known drug across all conditions. Compare the conditions and conditionMeshTerms against its original indication to surface off-label exploration.

curl "https://api.amass.tech/api/v1/cores/trialcore/records\
?query=metformin\
&studyType=INTERVENTIONAL\
&limit=200" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Then group results by conditionMeshTerms to see the full spread of indications being tested.

Example response (50 interventional metformin trials — 32 unique conditions, 19 diabetes-related, 46 other):

Conditions beyond diabetes:

Map Public Funding

Where is NIH investing in rare disease research?

Filter TrialCore by sponsorType: NIH for a condition area. The secondaryIds field often contains grant numbers, and facilityCountries shows where the funded research is happening.

curl "https://api.amass.tech/api/v1/cores/trialcore/records\
?query=rare+disease+gene+therapy\
&sponsorType=NIH\
&limit=200" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Example response (25 NIH-sponsored trials across 2 countries):

Geographic spread: US (17), PR (1), ML (1)

BiomedCore

Automate Literature Reviews

Build me an evidence table for CAR-T therapy in DLBCL.

Use BiomedCore search with date ranges and journal quality filters to retrieve and prioritize evidence. The publicationTypes field lets you programmatically filter for the study designs you care about (clinical trials, meta-analyses, randomized controlled trials). Request include=fulltext for downstream NLP processing.

curl "https://api.amass.tech/api/v1/cores/biomedcore/records\
?query=CAR-T+chimeric+antigen+receptor+DLBCL\
&minPublicationDate=2020-01-01\
&minJournalQualityJufo=1\
&include=fulltext\
&limit=200" \
  -H "Authorization: Bearer amass_YOUR_KEY"

What you get per record: title, abstract, fulltext, authors, publicationDate, citationCount, journalQualityJufo, publicationTypes, meshTerms, doi — structured for direct ingestion into your evidence pipeline.

Example response (25 papers since 2020, journal quality >= 1 — 24 with abstract, 9 with fulltext):

Gauge Scientific Maturity

Is there enough evidence behind PCSK9 inhibition to justify this investment?

Search BiomedCore for literature on a target, filtered to high-quality journals. The journalQualityJufo score (0–3, based on the Finnish JuFo classification) lets you quickly separate signal from noise. Combine with citation counts to find the foundational papers.

curl "https://api.amass.tech/api/v1/cores/biomedcore/records\
?query=PCSK9+inhibition+cardiovascular+outcomes\
&minJournalQualityJufo=2\
&minCitationCount=50\
&limit=50" \
  -H "Authorization: Bearer amass_YOUR_KEY"

What you get: Titles, abstracts, journal names, citation counts, publication types (meta-analysis, clinical trial, review), and MeSH terms — structured and ready for analysis.

Example response (25 high-impact papers, journal quality >= 2, citations >= 50):

DrugCore

Survey Drug Modalities

Which modalities are being pursued against this target, and how far have they advanced?

Search DrugCore for a drug class or indication and slice the results with the drugType and maxClinicalStage filters. One query separates the approved peptides from the small molecules still in Phase 1 and the antibodies in preclinical — a fast read on where the innovation (and the competition) sits. Search matches drug names, trade names, synonyms, and mechanism-of-action targets (target gene symbols, synonyms, names, and mechanism text), so you can query by drug-class terms (e.g. GLP-1) or by target gene symbol.

# All approved drugs in the class
curl "https://api.amass.tech/api/v1/cores/drugcore/records\
?query=GLP-1\
&maxClinicalStage=APPROVAL\
&limit=100" \
  -H "Authorization: Bearer amass_YOUR_KEY"

# Narrow to a single modality — e.g. proteins (peptides) only
curl "https://api.amass.tech/api/v1/cores/drugcore/records\
?query=GLP-1\
&drugType=PROTEIN\
&limit=100" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Each record carries name, tradeNames, drugType, maxClinicalStage, structure (inchiKey, canonicalSmiles), and mechanismsOfAction (pharmacological targets enriched with HGNC gene metadata) — enough to build a modality-vs-stage or modality-vs-target matrix without leaving DrugCore.

Example response — sweeping query=GLP-1 (100 records) and bucketing by drugType × maxClinicalStage:

To go one step further, request include=referencesTrialCore on the candidates and pull the trials directly from TrialCore to see who is running each program.

Trace Drug Relationships

What salts, prodrugs, and combination products relate to this molecule?

DrugCore organizes related records into a parent/child hierarchy. Request parent and children on a DrugCore record to walk that tree — from a base molecule down to its salt forms and fixed-dose combinations, or from a formulation back up to its parent active ingredient. These are intra-core links, so every ID stays within DrugCore (AMDC_...).

# Fetch a base molecule (atorvastatin) with its hierarchy links
curl "https://api.amass.tech/api/v1/cores/drugcore/records/AMDC_NmNGySqag6pdQd185ed33oG1X3D\
?include=parent&include=children" \
  -H "Authorization: Bearer amass_YOUR_KEY"
# Response includes:
#   "parent":   null                                    (top of the hierarchy)
#   "children": ["AMDC_MZuw8HPfCpN1OrzxasGKjVOsaja"]    (the salt form)

# ...walk down into the child form (atorvastatin calcium)
curl "https://api.amass.tech/api/v1/cores/drugcore/records/AMDC_MZuw8HPfCpN1OrzxasGKjVOsaja\
?include=parent" \
  -H "Authorization: Bearer amass_YOUR_KEY"
# "parent": "AMDC_NmNGySqag6pdQd185ed33oG1X3D"  — points back to atorvastatin

A null parent means the record sits at the top of its hierarchy; an empty children array means it is a leaf. Resolving the hierarchy lets you collapse salt forms and trade names onto a single active ingredient — useful when deduplicating a pipeline or rolling up trial and literature evidence across every form of the same drug.

Example response — the atorvastatin hierarchy:

GeneCore

Prioritize Targets By Constraint

Which genes in a pathway are highly essential for survival, yet heavily constrained against mutation?

Search GeneCore for a specific gene family, pathway, or concept and filter the results using gnomAD genetic constraint (maxConstraintLoeuf) and CRISPR essentiality (isEssential). Setting a low loss-of-function observed/expected upper-bound fraction (LOEUF) threshold (e.g. <= 0.6 for gnomAD v4.0) isolates genes that are highly intolerant to loss-of-function variation. Pairing this with isEssential=true filters down to targets where cellular viability is highly dependent on their function—ideal for finding critical cancer dependencies or target-vulnerability nodes.

# Search for kinases that are highly essential and heavily constrained
curl "https://api.amass.tech/api/v1/cores/genecore/records\
?query=kinase\
&isEssential=true\
&maxConstraintLoeuf=0.6\
&limit=5" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Each record carries HGNC identifiers, NCBI RefSeq summaries, and deep target intelligence. The gnomadConstraint.lossOfFunction.loeuf provides the exact LOEUF value, while depmapEssentiality details the cell lines most dependent on the gene (selective dependency).

To evaluate structural druggability and localized cellular context, opt-in to fetch Swiss-Prot protein details using include=protein:

# Fetch structural and localization details for a prioritized target
curl "https://api.amass.tech/api/v1/cores/genecore/records/AMGC_MTOR\
?include=protein" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Example response — sweeping query=kinase (with isEssential=true and maxConstraintLoeuf=0.6):

Resolving the prioritized gene IDs with include=protein reveals their biochemical features and structural availability:

RegulatoryCore

Compare US and EU Approvals

Is this drug approved in both the US and the EU — and does its status differ between markets?

Search RegulatoryCore across both agencies and read the authorizationsByAgency field on every record. It is always populated and lists the same product's authorizations in the other market — each with its own authorizationStatus — so a status divergence (active in the US, withdrawn in the EU) reads straight off a single response, no second query required.

curl "https://api.amass.tech/api/v1/cores/regulatorycore/records\
?query=pembrolizumab\
&agency=FDA\
&agency=EMA\
&limit=50" \
  -H "Authorization: Bearer amass_YOUR_KEY"

What you get: agency, name, authorizationStatus, authorizationDate, marketingAuthorisationHolder, and authorizationsByAgency — the cross-market link carrying each counterpart authorization's agency, name, and authorizationStatus.

Example response (cross-market status comparison, one row per product):

Pull the full picture for a single product by following the linked Amass ID:

# authorizationsByAgency on the FDA record points to the EMA record's AMRC_ ID
curl "https://api.amass.tech/api/v1/cores/regulatorycore/records/AMRC_eu123\
?include=emaDetails&include=fdaDetails" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Track Regulatory Designations

Which oncology biologics reached the market on an accelerated or breakthrough pathway?

Filter RegulatoryCore with hasDesignation to surface authorizations granted an expedited program, then read the designations array to see how each agency-native program maps onto a shared comparison axis (REVIEW_ACCELERATION, DEVELOPMENT_SUPPORT, EARLY_ACCESS_BASIS). Because hasDesignation only applies to the agency that owns each program, you can list FDA-only and EMA-only designations together and keep records from both markets — each side matched on its own column.

curl "https://api.amass.tech/api/v1/cores/regulatorycore/records\
?query=oncology\
&agency=FDA\
&agency=EMA\
&moleculeType=ANTIBODY\
&moleculeType=ANTIBODY_DRUG_CONJUGATE\
&hasDesignation=BREAKTHROUGH_THERAPY\
&hasDesignation=ACCELERATED_APPROVAL\
&hasDesignation=PRIME\
&hasDesignation=CONDITIONAL_MA\
&limit=100" \
  -H "Authorization: Bearer amass_YOUR_KEY"

What you get: name, agency, authorizationStatus, and the designations array — each entry carrying axis, type, nativeName, basis, and postMarketingObligation.

Example response (designations grouped by shared axis):

For records with an EARLY_ACCESS_BASIS row, the basis field tells you why the early authorization was granted — SURROGATE_ENDPOINT, INCOMPLETE_DATA, or UNCONFIRMABLE_DATA — which is often the first thing a reviewer wants to see.

Map Orphan Drug Approvals

Which orphan-designated drugs have been authorized in the EU since 2020 — and are they still active?

isOrphan is an exact cross-walk between the FDA Orphan Drug and EMA Orphan Medicine designations, so a single filter works across both agencies. Combine it with an agency filter, an authorization-status filter, and a date range to scope a rare-disease landscape in one query.

curl "https://api.amass.tech/api/v1/cores/regulatorycore/records\
?query=rare+disease\
&agency=EMA\
&isOrphan=true\
&authorizationStatus=ACTIVE\
&authorizationStatus=CONDITIONAL\
&minAuthorizationDate=2020-01-01" \
  -H "Authorization: Bearer amass_YOUR_KEY"

What you get: name, activeSubstance, moleculeType, authorizationStatus, authorizationDate, marketingAuthorisationHolder, and isOrphan — everything needed to build a rare-disease approval table.

Example response (active or conditionally authorized EU orphan medicines since 2020):

To compare orphan approvals across both markets at once, drop the agency filter (or list both agency=FDA&agency=EMA) and read authorizationsByAgency on each record to see whether the same product carries orphan status in the other market too.

Search the Regulatory Source Documents

Which approved drugs actually mention this clinical phrase — in the warnings, the mechanism, the trial readout — even when it never appears in the structured fields?

Structured metadata (name, active substance, indication, holder) only scratches the surface of a regulatory file. The substance lives in the prose: the boxed warnings, the immune-mediated reactions, the surrogate endpoints that justified an accelerated approval, the drug-interaction tables. RegulatoryCore indexes the parsed full text of every FDA label, FDA review, EMA SmPC, and EMA EPAR, so a plain query= sweeps all of it.

When a phrase matches document content rather than metadata, each record comes back with a documentSections[] array — the exact sections that hit, each with a matchedText excerpt and an addressable documentSectionId.

Mine a safety signal across the whole corpus

Find every authorization whose documents discuss a specific adverse reaction — the backbone of a fast pharmacovigilance scan.

# Every drug whose label/review/SmPC/EPAR warns about immune-mediated hepatitis
curl "https://api.amass.tech/api/v1/cores/regulatorycore/records\
?query=immune-mediated%20hepatitis\
&limit=25" \
  -H "Authorization: Bearer amass_YOUR_KEY"

This surfaces the checkpoint-inhibitor class (Keytruda, Opdivo, Tecentriq, …) even though "hepatitis" is nowhere in their structured indication — it lives in section 5 (Warnings and Precautions) of each label.

What you get on each record, alongside the usual metadata:

{
  "amassId": "AMRC_...",
  "name": "Keytruda",
  "documentSections": [
    {
      "documentSectionId": "AMRCDS_...",
      "docType": "FDA_LABEL",
      "path": "5.2",
      "title": "Immune-Mediated Hepatitis",
      "matchedText": "...KEYTRUDA can cause immune-mediated hepatitis. Monitor patients for changes in liver function..."
    }
  ]
}

More phrases worth sweeping

The same pattern answers a wide range of questions that metadata filters cannot:

# Cardiac liability buried in the warnings: QT prolongation / torsades
curl "https://api.amass.tech/api/v1/cores/regulatorycore/records?query=torsades%20de%20pointes" \
  -H "Authorization: Bearer amass_YOUR_KEY"

# Accelerated-approval reasoning in FDA review text: surrogate endpoints
curl "https://api.amass.tech/api/v1/cores/regulatorycore/records?query=progression-free%20survival%20surrogate%20endpoint" \
  -H "Authorization: Bearer amass_YOUR_KEY"

# Rare but serious class warning: PML in immunomodulators
curl "https://api.amass.tech/api/v1/cores/regulatorycore/records?query=progressive%20multifocal%20leukoencephalopathy" \
  -H "Authorization: Bearer amass_YOUR_KEY"

# Formulation / excipient sensitivity across SmPCs
curl "https://api.amass.tech/api/v1/cores/regulatorycore/records?query=contains%20polysorbate%2080" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Read the full section

matchedText is a snippet, not the whole section. Follow any documentSectionId to pull the complete text:

curl "https://api.amass.tech/api/v1/cores/regulatorycore/records/AMRC_abc123\
/document-sections/AMRCDS_def456" \
  -H "Authorization: Bearer amass_YOUR_KEY"
# → { "data": { ..., "content": "full section text", "sourceUrl": "...label.pdf" } }

How matching works. A single query matches structured metadata and sweeps document content in one call. A record with an empty documentSections[] matched on metadata only. Section path is numbered and stable for FDA labels and EMA SmPCs but opaque for FDA reviews and EMA EPARs — always address a section by its documentSectionId, not its path.

Cross-Core

Connect Trials To Evidence

What published evidence supports this company's lead program?

Request include=referencesBiomedCore on any TrialCore record to get the Amass IDs of the publications cited by the trial. Fetch those directly via BiomedCore to build a complete picture: the trial's design and the science it's built on.

The array is ordered results-first. Publications that register this trial's id in their own PubMed record come first, then the background literature, so the leading entries are the trial's own results papers where they exist.

# 1. Find the trial and request referenced publication IDs
curl "https://api.amass.tech/api/v1/cores/trialcore/records\
?query=semaglutide+NASH\
&phase=PHASE3\
&include=referencesBiomedCore" \
  -H "Authorization: Bearer amass_YOUR_KEY"

# Response includes: "referencesBiomedCore": ["AMBC_abc123", "AMBC_def456"]

# 2. Fetch each referenced publication by its Amass ID
curl "https://api.amass.tech/api/v1/cores/biomedcore/records/AMBC_abc123" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Example response — 25 trials found, 54 unique referenced publications across 15 trials:

Resolve those IDs via BiomedCore to get the full publications:

Find Investigators And Sites

Who are the active researchers in gene therapy for sickle cell disease, and where are they?

Combine BiomedCore author search with TrialCore geographic data:

• BiomedCore with include=authorsMetadata gives you authors with ORCIDs, institutional affiliations, ROR identifiers, and country codes
• TrialCore facilityCountries shows where trials are physically running

curl "https://api.amass.tech/api/v1/cores/biomedcore/records\
?query=gene+therapy+sickle+cell+disease\
&minCitationCount=10\
&minPublicationDate=2022-01-01\
&include=authorsMetadata\
&limit=100" \
  -H "Authorization: Bearer amass_YOUR_KEY"

Example response (288 unique authors across 25 papers, cited >= 10, since 2022):

Map Drug Evidence Base

Given a drug, what clinical trials and published science back it?

Start from a known drug in DrugCore and request both cross-core links — referencesTrialCore and referencesBiomedCore — to fan out from a single molecule to every associated clinical trial (TrialCore) and publication (BiomedCore) in one call. If you only have a ChEMBL ID, resolve it to an Amass ID first via the lookup endpoint.

# 1. (Optional) Resolve a ChEMBL ID to an Amass ID — metformin is CHEMBL1431
curl -X POST "https://api.amass.tech/api/v1/cores/drugcore/records/lookup" \
  -H "Authorization: Bearer amass_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"items": [{"chemblId": "CHEMBL1431"}]}'
# Response includes: "amassIds": ["AMDC_X76106vVMepKB3UsB3huC0mPZEu"]

# 2. Fetch the drug with both cross-core link sets
curl "https://api.amass.tech/api/v1/cores/drugcore/records/AMDC_X76106vVMepKB3UsB3huC0mPZEu\
?include=referencesTrialCore&include=referencesBiomedCore" \
  -H "Authorization: Bearer amass_YOUR_KEY"
# Response includes (metformin):
#   "referencesTrialCore":   ["AMTC_3vmicVPly4xk2vr9yCofBt2gtKh", ...]   ← 1,818 IDs
#   "referencesBiomedCore":  ["AMBC_SIO2LpmEXv62nEIvkBWxLZiskfT", ...]   ← 4 IDs

# 3. Resolve any linked ID directly in its Core
curl "https://api.amass.tech/api/v1/cores/trialcore/records/AMTC_3vmicVPly4xk2vr9yCofBt2gtKh" \
  -H "Authorization: Bearer amass_YOUR_KEY"
curl "https://api.amass.tech/api/v1/cores/biomedcore/records/AMBC_SIO2LpmEXv62nEIvkBWxLZiskfT" \
  -H "Authorization: Bearer amass_YOUR_KEY"

From one drug record you get a ready-made list of trial and publication Amass IDs — no keyword matching, no name disambiguation. Resolve the trial IDs to read phase, sponsor, and outcomes, and the publication IDs to read titles, journals, and citation counts.

Note on coverage. referencesTrialCore is densely populated (metformin links to 1,818 trials), while referencesBiomedCore is currently sparse (single digits, sometimes empty — e.g. pembrolizumab has 2,408 trial links but no biomed links yet). Always check the array length before relying on it, and treat an empty list as "no links recorded," not "no evidence."

Link Targets To Pharmacology

What clinical-stage or approved drugs exist that target this specific gene, and what are their structures?

Query GeneCore for a target or gene family and request the cross-core link referencesDrugCore. This pulls down the Amass IDs of all drugs and clinical candidates in DrugCore that modulate that gene. By following this cross-core link, you can immediately map biological targets to their active drug programs and structural molecules in one seamless workflow.

# 1. Search for a target gene and request its linked DrugCore records
curl "https://api.amass.tech/api/v1/cores/genecore/records\
?query=EGFR\
&geneType=PROTEIN_CODING\
&include=referencesDrugCore\
&limit=1" \
  -H "Authorization: Bearer amass_YOUR_KEY"

# Response includes:
# {
#   "data": [
#     {
#       "amassId": "AMGC_01EGFR...",
#       "symbol": "EGFR",
#       "referencesDrugCore": ["AMDC_osimertinib...", "AMDC_gefitinib...", "AMDC_erlotinib..."]
#     }
#   ]
# }

# 2. Fetch the corresponding DrugCore records to extract their structures, clinical stages, and types
curl "https://api.amass.tech/api/v1/cores/drugcore/records/AMDC_osimertinib..." \
  -H "Authorization: Bearer amass_YOUR_KEY"

From a gene search, you transition directly to structural pharmacology. By fetching the linked DrugCore IDs, you get chemical data (canonicalSmiles, inchiKey), drug class info, and its maximum regulatory clinical stage.

Example response — searching query=EGFR with include=referencesDrugCore and resolving the linked drugs:

Cascading Further. Once you have resolved the DrugCore IDs, you can continue fanning out across other cores. Request referencesTrialCore on any of these molecules to see active clinical trial protocols in TrialCore, or fetch their regulatory approvals in RegulatoryCore to inspect FDA/EMA labels.