An implementation of ESM2 in Equinox+JAX

Repository of the paper "Exploring sequence landscape of biosynthetic gene clusters with protein language models" published at ICML2024 workshop Machine Learning for Life and Material Science: From Theory to Industry applications

Prediction of anti-fungal proteins using protein language models

Residue-residue contact prediction using ESM2 enhanced with template-based structural priors from homologous sequences.

Real-time, zero-shot Variant Effect Prediction (VEP) and in-silico mutagenesis engine powered by Meta's ESM-2. Achieves sub-second inference on edge and consumer hardware.

Zero-shot variant pathogenicity prediction via spectral covariance analysis of ESM2 hidden states | Claw4S 2026

🧬 CAFA6 Protein Function Prediction | Hybrid DL + BLAST Ensemble | Solo Participant | Public LB 0.283

Federated learning framework for protein language models (ESM2) using NVFlare + Flower. Supports multi-GPU simulation, HPC/SLURM deployment, and HuggingFace dataset integration.

Similarity search for protein sequences using ESM-2 embeddings and Approximate Nearest Neighbor (ANN) methods.

Fusion model of ProtBERT and ESM-2 for cell-penetrating peptide prediction (Reproduction of FusPB-ESM2, Comp. Biol. Chem., 2024)

Reproducible Nextflow + Python pipeline for TCR-epitope binding prediction with Bayesian calibration

Protein subcellular localization classifier using ESM-2 embeddings and PyTorch MLP.

Exploring the AMR protein embedding landscape with ESM2 and Ankh. Interactive tool at HuggingFace Spaces.

AI-powered TCR pathology classification using ESM2 embeddings for disease-reactive TCR discovery

MVP de cribado virtual asistido por IA y docking molecular con biblioteca botanica de Ecuador y Amazonia.

ESM2-15B Protein Embeddings

Multi-phase peptide ML pipeline for hemolysis prediction, generative design, and closed-loop optimization, built on transformers and modern PyTorch tooling.

TCR Epitope Generation Model with Top-K Prediction

A machine learning pipeline for eukaryotic signal peptide prediction — from UniProtKB data collection to a CNN-LSTM classifier on ESM-2 embeddings. Implements and benchmarks three methods: a von Heijne PSWM baseline, an SVM with biochemical features, and a deep learning model (F1=0.962, ROC-AUC=0.998).