Skip to main content
Back to challenges
Challenge

Develop a Liquid Biopsy Variant Caller for Clinical Diagnostics with DeepVariant and FHIR

Clinical deployment of liquid biopsy screening requires high sensitivity variant detection followed by standardized electronic health record integration. Build an automated end-to-end pipeline using DeepVariant models and HL7 FHIR APIs to transform low-frequency cfDNA variants into compliant FHIR DiagnosticReport payloads with 99% precision.

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.

End-to-end genomic diagnostic pipeline converting DeepVariant liquid biopsy variant calls into HL7 FHIR DiagnosticReport objects.

Delivery guide

How work is evaluated

Evaluation

Evaluates low-frequency variant call precision and FHIR payload schema validation.

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

fhir_conformance_check

Ensures output is valid FHIR DiagnosticReport resource.

Binary check

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

Dimension 2

variant_calling_precision

Precision of detected low-VAF somatic variants. • target: 0.99 • 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

  • Process cfDNA BAM alignment files using Google DeepVariant variant calling pipeline.

  • Filter low allele frequency (<1% VAF) somatic tumor mutations from sequencing noise.

  • Format identified variants into standardized HL7 FHIR DiagnosticReport and Observation resources.

Resources and assets

Reference links and supporting material

Dataset notes

Dataset of synthetic cfDNA VCF records containing low-frequency somatic tumor variants (0.1% to 2% VAF).

Challenge setup

Evaluates low-frequency variant call precision and FHIR payload schema validation.

Challenge input

JSON containing variant_vcf_record and patient_id

DeepVariant

Deep learning variant caller for NGS sequence data.

FHIR

HL7 standard for healthcare data interoperability.

Evaluated output

JSON compliant with FHIR DiagnosticReport resource schema

Evaluation checks
  • Ensures output is valid FHIR DiagnosticReport resource.
  • Precision of detected low-VAF somatic variants. • target: 0.99 • range: 0-1
Configured evaluation evidence
  • Benchmark: FHIRGeno-Bench
  • Variant Calling Precision target: 0.99
  • 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
  • Agno · Required
  • Zed · Optional
  • Command R+ · Optional
Orchestration
  • Agno · Required

Evaluation contract

  • fhir_conformance_check · Weight 1
  • variant_calling_precision · Weight 1

Metrics

  • variant_calling_precision · Target 0.99

Benchmarks

  • FHIRGeno-Bench
  • biotech

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing variant_vcf_record and patient_id

    Output: JSON compliant with FHIR DiagnosticReport resource schema

    Execution harness: Python

Evaluation tools

  • DeepVariant · variant_caller
  • FHIR · data_standard

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Develop a Liquid Biopsy Variant Caller for Clinical Diagnostics with DeepVariant and FHIR