Reproducible biosecurity compendium: ESM-2 homology-cliff failure analysis, pre-registered

This paper has been published on Interdisciplinary Sciences: Computational Life Sciences.

Code for processing FASTA files into batched, tokenized, and padded sequences, and training FlashAttention-based implementations of ESM-2

A Graph Neural Network (Siamese GATv2) pipeline using ESM-2 embeddings to map the Dengue NS1–Human interactome. Discovered a novel link between NS1 and Carbonic Anhydrases.

Graph-based Protein-Peptide Interaction Predictor. Reranking candidates using ESM-2 embeddings and Graph Neural Networks (GNN) on PROPEDIA data.

Project on fine tuning ESM models for various purposes

Bioinformatics capstone of the UoA DI 'Software Development for Algorithmic Problems' trilogy. Protein remote-homolog search on SwissProt-50k via ESM-2 embeddings + 6 ANN methods (brute, LSH, Hypercube, IVF-Flat, IVF-PQ, Neural LSH), benchmarked against BLAST.

Programmable genome‑writing knowledge graph and GNN toolkit for DNA‑modifying enzymes, combining CRISPR/CAST/Fanzor annotations with ESM‑2 embeddings, GraphSAGE/GAT link prediction, and an editor‑selection API for gene‑editing use cases.

A hybrid C++/Python pipeline for remote protein homology detection, coupling ESM-2 language model embeddings with custom Neural-LSH for scalable approximate nearest neighbor search.

ML classifier for genome-editing enzymes, assigning nuclease/recombinase/transposase classes that predicts mechanism class and flags IS110-family composite recombinases overlooked by existing tools.

Empirical dual-use risk assessment of protein language models (ESM-2) and structure-based design tools (ProteinMPNN)

ProteinFlex is a comprehensive platform for protein structure analysis and drug discovery, leveraging advanced AI and machine learning techniques. The platform combines state-of-the-art protein structure prediction with interactive visualization and sophisticated drug discovery tools.