NtiriLabScience · AI · Publishing
NtiriLab platform

Arabidopsis Stress Protein Predictor

A machine-learning project exploring protein-sequence representations and stress-related protein prediction.

The problem

Protein function and stress-response annotation require evidence-rich workflows that can combine biological knowledge with high-dimensional sequence features.

The approach

Compare engineered amino-acid features with ESM2 protein embeddings across multiple classifiers, cross-validation strategies and calibrated thresholds.

Evidence and outputs

  • Model comparison
  • Feature pipeline
  • Interactive prototype
  • Evaluation notes

Tools and methods

Pythonscikit-learnESM2StreamlitPandas

Evidence standard

This profile deliberately separates current status from ambition. Links appear only when a repository, demo or public output has been supplied.