What problem does it solve?
Proteomic biological age clocks normally return a single black-box number per organ, making it impossible to explain which proteins drive the prediction. This Skill breaks every Goeminne organ-aging clock prediction down into individual protein contributions (NPX × coefficient) so results are inspectable and explainable.
Core Features & Use Cases
- Per-protein contribution breakdown: Ranks proteins by absolute coefficient and reports each protein's NPX value, coefficient, and contribution to predicted organ age.
- Multi-organ and multi-generation support: Runs any organ from Goeminne et al. (2025), with gen1 chronological-age models or gen2 mortality-hazard models converted to years via the Gompertz transform.
- Protein filtering for demos:
--top-n and --min-abs-coef flags let you show only the strongest drivers for a single sample.
- Structured outputs: Writes a Markdown report, machine-readable JSON, a long-format protein_contributions.csv table, and a replayable commands.sh.
- Use Case: A researcher has an Olink NPX table for one patient and wants to explain why the Heart clock predicts accelerated aging. They run the skill with
--organs Heart --sample-id PATIENT_001 --top-n 5 and get a ranked list of the five proteins driving that prediction.
Quick Start
Ask the agent to run the organ aging studio demo on the bundled synthetic Olink data and show which proteins drive the Heart and Brain age predictions for sample DEMO_000.
