tooluniverse-immunotherapy-response-prediction

What it does

Predict patient response to immune checkpoint inhibitors (ICIs) using multi-biomarker integration. Transforms a patient tumor profile (cancer type + mutations + biomarkers) into a quantitative ICI Response Score with drug-specific recommendations, resistance risk assessment, and monitoring plan.

KEY PRINCIPLES:

1. Report-first approach - Create report file FIRST, then populate progressively
2. Evidence-graded - Every finding has an evidence tier (T1-T4)
3. Quantitative output - ICI Response Score (0-100) with transparent component breakdown
4. Cancer-specific - All thresholds and predictions are cancer-type adjusted
5. Multi-biomarker - Integrate TMB + MSI + PD-L1 + neoantigen + mutations
6. Resistance-aware - Always check for known resistance mutations (STK11, PTEN, JAK1/2, B2M)
7. Drug-specific - Recommend specific ICI agents with evidence
8. Source-referenced - Every statement cites the tool/database source
9. English-first queries - Always use English terms in tool calls

How it works

The agent is instructed to process input cancer type, mutations, and optional biomarkers to generate a structured, evidence-graded report across 11 phases. It starts by standardizing inputs (cancer context, mutations, gene IDs), then conducts phase-specific analyses (TMB classification, neoantigen burden, MSI/MMR status, PD-L1 evaluation, immune microenvironment, mutation-based predictors, clinical evidence, and multi-biomarker score integration). It computes an ICI Response Score and provides drug-specific recommendations with evidence, resistance risk factors, and a monitoring plan. The workflow includes phase-specific steps such as resolving cancer type to an EFO ID, parsing mutations, identifying driver and resistance mutations, estimating neoantigen load, and retrieving FDA-approved ICIs and clinical evidence. The system uses the verified tool parameter reference before tool calls and relies on cancer-type-adjusted thresholds and FDA contexts for decisions.

Concrete steps include:

• Resolve cancer type to EFO ID and fetch cancer-specific ICI context.
• Parse and normalize mutations (gene, variant, type).
• Map genes to IDs using gene-query tools.
• Classify TMB if provided, or estimate from mutation list, then apply cancer-specific thresholds.
• Estimate neoantigen load from mutation counts with defined conversion rules.
• Assess MSI/MMR status, check MMR genes in mutations, and consult FDA MSI-H approvals.
• Classify PD-L1 expression with cancer-specific thresholds and FDA cutoffs.
• Profile immune microenvironment and assess immune escape indicators.
• Identify resistance/sensitivity mutations (e.g., STK11, PTEN, JAK1/2, B2M; POLE).
• Retrieve FDA-approved ICIs and clinical evidence for the cancer type.
• Integrate multi-biomarker data into a single ICI Response Score (0-100) with component breakdown.
• Provide clinical recommendations for ICI drugs with a monitoring plan and alternative strategies.

When to use it

Apply when user asks:

• a question about likely response to immunotherapy, selecting an ICI, or biomarker-guided ICI decisions across cancer types.
• scenarios involving combinations or cancer-type–specific ICI choices with biomarker considerations.

What it can touch

It uses tools to obtain disease descriptions, drug descriptions, associated drugs, drug mechanisms, and gene information such as:

• OpenTargets_get_disease_id_description_by_name
• OpenTargets_get_drug_id_description_by_name
• OpenTargets_get_associated_drugs_by_disease_efoId
• OpenTargets_get_drug_mechanisms_of_action_by_chemblId
• OpenTargets_get_approved_indications_by_drug_chemblId
• OpenTargets_get_drug_description_by_chemblId
• OpenTargets_get_associated_targets_by_drug_chemblId
• MyGene_query_genes
• EnsemblVEP_annotate_rsid (for variant impact)
• civic and IEDB related epitopes data
• UniProt_get_function_by_accession

Caveats

• The approach relies on cancer-type–specific thresholds and FDA contexts; results are evidence-graded with tiers (T1-T4) and should be interpreted in clinical context.
• It flags known resistance mutations (STK11, PTEN, JAK1/2, B2M) and considers mutation-based resistance/sensitivity signals; actual clinical decisions depend on full patient data and clinical judgment.
• No outcomes are promised beyond the explicit scoring, thresholds, and recommendations described in the workflow.