Autism PPI Network Perturbation Analysis Agent with Google ADK
Understanding how autism mutations disrupt protein-protein interaction networks requires multi-omic graph analysis. Build a Google ADK agent hosted on Baseten to analyze PPI perturbation scores and report network centrality changes across thousands of neurodevelopmental gene variants.
What you are building
Develop a protein interaction network analysis agent using Google ADK deployed via Baseten to compute topological graph disruptions.
How work is evaluated
Evaluates graph perturbation scoring accuracy and output formatting.
Datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
Centrality Delta Range Check
Ensures calculated deltas fall within [-1.0, 1.0].
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Graph Metric Mean Absolute Error
MAE between computed centrality changes and expected network values. • target: 0.01 • range: 0-0.5
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Learning goals
Construct multimodal graph agents using Google Agent Development Kit (ADK).
Deploy custom low-latency inference functions on Baseten platform.
Analyze network topology changes (betweenness centrality, degree centrality) induced by missense mutations.
Generate interactome perturbation summary matrices.
Resources
Protein-protein interaction graph dataset containing wildtype interactomes and mutant edge loss annotations for neurodevelopmental genes.
Challenge setup
Evaluates graph perturbation scoring accuracy and output formatting.
Challenge input
JSON with mutated gene, wildtype PPI edge list, and altered binding edges
Google ADK
Provides structure for building multimodal analytical agents.
Baseten
Delivers fast microservice deployment for graph computation.
Evaluated output
JSON with delta_degree_centrality, delta_betweenness, and hub_status
- Ensures calculated deltas fall within [-1.0, 1.0].
- MAE between computed centrality changes and expected network values. • target: 0.01 • range: 0-0.5
- Benchmark: ProteomicNetwork-Bench
- Graph Metric Mean Absolute Error target: 0.01
- 1 public reference case
- Python execution harness
- Python sandbox (unavailable on Versalist)
Shows the first task, up to three tools, and up to three evidence items.
View technical recipe
Configured tools
- Google · Required
- Google ADK · Optional
- Baseten · Optional
Evaluation contract
- Centrality Delta Range Check · Weight 1
- Graph Metric Mean Absolute Error · Weight 1
Metrics
- Graph Metric Mean Absolute Error · Target 0.01
Benchmarks
- ProteomicNetwork-Bench
- biotech
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON with mutated gene, wildtype PPI edge list, and altered binding edges
Output: JSON with delta_degree_centrality, delta_betweenness, and hub_status
Execution harness: Python
Evaluation tools
- Google ADK · Agent Framework
- Baseten · Inference Host
Recipe freeze marker
No recipe freeze time is recorded.