Skip to main content
Back to challenges
Challenge

Construct a Cell-Free DNA Toxicity Monitoring RAG with LlamaIndex

Monitoring adverse drug reactions during clinical trials requires rapid retrieval across extensive cell-free DNA (cfDNA) epigenetic fragmentation studies. Build an advanced RAG querying engine using LlamaIndex to query drug safety assays and extract target organ toxicity signals with 94% recall.

Data ScienceHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Build an advanced RAG vector index using LlamaIndex to retrieve cell-free DNA drug toxicity biomarkers.

Delivery guide

How work is evaluated

Evaluation

Evaluates retrieval recall and entity accuracy for cfDNA organ toxicity queries.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

organ_match_test

Verifies correct target organ identified.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

retrieval_recall

Recall metric of target ground truth documents. • target: 0.94 • range: 0-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Implement VectorStoreIndex and ContextChatEngine using LlamaIndex.

  • Index literature and trial reports covering cfDNA genomic safety monitoring.

  • Retrieve precise organ-specific methylation and fragmentation patterns linked to drug toxicity.

Resources and assets

Reference links and supporting material

Dataset notes

Curated corpus of 50 scientific papers and safety reports on cell-free DNA methylation and fragmentation markers of organ toxicity.

Challenge setup

Evaluates retrieval recall and entity accuracy for cfDNA organ toxicity queries.

Challenge input

JSON containing query text

LlamaIndex

Framework specified for indexing and querying complex documents.

Evaluated output

JSON with detected_organ_toxicity, target_biomarker, and recall_score

Evaluation checks
  • Verifies correct target organ identified.
  • Recall metric of target ground truth documents. • target: 0.94 • range: 0-1
Configured evaluation evidence
  • Benchmark: BioRAG-Bench
  • Retrieval Recall target: 0.94
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • Llama Index · Required
  • Softgen AI · Optional
  • Langbase · Optional

Evaluation contract

  • organ_match_test · Weight 1
  • retrieval_recall · Weight 1

Metrics

  • retrieval_recall · Target 0.94

Benchmarks

  • BioRAG-Bench
  • biotech

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing query text

    Output: JSON with detected_organ_toxicity, target_biomarker, and recall_score

    Execution harness: Python

Evaluation tools

  • LlamaIndex · agent_framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Construct a Cell-Free DNA Toxicity Monitoring RAG with LlamaIndex