Challenge

Protein Functional Annotation Pipeline

Build a high-performance protein annotation agent using the Google ADK (Agent Development Kit). The agent's core responsibility is to reconcile unreviewed protein sequences with reviewed UniProtKB entries. You will deploy a sequence alignment service on Modal to handle heavy BLAST-like computations or HMMER searches. The Google ADK agent will orchestrate the flow: receiving an unreviewed sequence, triggering a Modal inference job for alignment, and then fetching metadata from UniProt (taxonomy, active sites, PTMs) to explain the functional confidence. The agent must handle the multimodal aspect of protein data (sequences and structural metadata) and return a detailed 'Evidence Reconciliation' report that identifies conserved domains and functional residues.

Data ScienceHosted by Vera
Status
Always open
Difficulty
Advanced
Points
500
Challenge brief

What you are building

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

Build a high-performance protein annotation agent using the Google ADK (Agent Development Kit). The agent's core responsibility is to reconcile unreviewed protein sequences with reviewed UniProtKB entries. You will deploy a sequence alignment service on Modal to handle heavy BLAST-like computations or HMMER searches. The Google ADK agent will orchestrate the flow: receiving an unreviewed sequence, triggering a Modal inference job for alignment, and then fetching metadata from UniProt (taxonomy, active sites, PTMs) to explain the functional confidence. The agent must handle the multimodal aspect of protein data (sequences and structural metadata) and return a detailed 'Evidence Reconciliation' report that identifies conserved domains and functional residues.

Datasets

Shared data for this challenge

Review public datasets and any private uploads tied to your build.

Loading datasets...
Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks, how much each dimension matters, and which criteria separate a passable run from a strong one.

Max Score: 2
Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1modal_execution

Modal Execution

Verifies that the alignment was actually performed on Modal serverless.

binary
Weight: 1
Binary check

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

Dimension 2annotation_accuracy

Annotation Accuracy

Percentage of correctly identified functional residues. • target: 0.95 • range: 0-1

binary
Weight: 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

  • Initialize Google ADK with Gemini 1.5 Pro to reason over protein sequence motifs and literature

  • Implement a 'modal.Function' that wraps Biopython's Blast/Align tools for serverless execution

  • Design an agent tool that queries the UniProt REST API using specific accessions and returns XML/JSON metadata

  • Build a logic gate that assigns confidence levels (Gold, Silver, Bronze) based on the 'Protein Existence' (PE) level in UniProt

  • Utilize Google ADK's Vertex AI integration to store and retrieve sequence embeddings for similarity search

  • Programmatically extract 'Features' (FT) from UniProt records to identify active site residues

  • Create a final reporting tool that generates a Markdown summary of the protein's predicted function with alignment visualizations

Start from your terminal
$npx -y @versalist/cli start protein-functional-annotation-pipeline

[ok] Wrote CHALLENGE.md

[ok] Wrote .versalist.json

[ok] Wrote eval/examples.json

Requires VERSALIST_API_KEY. Works with any MCP-aware editor.

Docs
Manage API keys
Host and timing
Vera

AI Research & Mentorship

Starts Available now
Evergreen challenge
Your progress

Participation status

You haven't started this challenge yet

Timeline and host

Operating window

Key dates and the organization behind this challenge.

Start date
Available now
Run mode
Evergreen challenge
Explore

Find another challenge

Jump to a random challenge when you want a fresh benchmark or a different problem space.

Useful when you want to pressure-test your workflow on a new dataset, new constraints, or a new evaluation rubric.

Tool Space Recipe

Draft
Environment
ModalServerless GPU runtime for containerized AI workloads.
Action Space
GoogleGoogle AI model provider
required
Google ADKAgent Development Kit for
Policy Serving
ModalServerless GPU runtime for containerized AI workloads.
Training Infra
ModalServerless GPU runtime for containerized AI workloads.
Evaluation
Rubric: 2 dimensions
·Modal Execution(1%)
·Annotation Accuracy(1%)
Gold items: 1 (1 public)

Frequently Asked Questions about Protein Functional Annotation Pipeline