CLE Peptide Discovery Pipeline

An AI-driven pipeline for the identification of CLE signaling peptides in plant proteomes using protein language model embeddings.

Overview

This pipeline leverages state-of-the-art Protein Language Models (ESM2 and ProtT5) to discover novel CLE peptides directly from plant proteomes. By coupling evolutionary-scale sequence embeddings with unsupervised clustering and supervised machine learning, this dual-model approach captures deep semantic features of the CLE family that escape traditional sequence alignment methods.

Pipeline Architecture

• Step 1: Embedding extraction (ESM2 + ProtT5):

  • Embeddings_ESM2.py
  • Embeddings_T5.py

• Step 2: Clustering analysis:

  • Cluster_maps.py

• Step 3: XGBoost training:

  • XGB_training.py

• Step 4: XGBoost cluster prediction:

  • Cluster_prediction.py

• Step 5: Sequence extraction:

  • Sequence_MEME_analysis.py
  • extract_all_candidates.py (candidates ≥ 0.5) → optional

Visualization

• Bokeh_visualization_positives.py