[ICML 2025] Fast and Low-Cost Genomic Foundation Models via Outlier Removal.

Assessing LLMs—AAM, DNABERT, DNABERT-2, GROVER—to improve the prediction of COVID-19 status ("Positive" or "Not detected") using microbiome data

Interpretable Deep Learning and Ensemble Models for Predicting Multidrug Resistance in Klebsiella pneumoniae

CRISPGen: An end-to-end conditional latent diffusion and dual-critic reinforcement learning framework for the de novo synthesis of high-efficiency, clinical-grade CRISPR/Cas9 guide RNAs with 99.7% off-target risk suppression.

LaTeX source and PDFs of the HantaBERT paper (English & Indonesian) — multi-task hantavirus classification with DNABERT-2.

FastAPI inference service for HantaBERT — classifies hantavirus nucleotide sequences by species, host, and geographic origin.

Multi-task Orthohantavirus classification by fine-tuning DNABERT-2 — species, host, and geographic origin in a single forward pass.

This project implements a hybrid machine learning approach for classifying breast cancer from DNA sequences using bidirectional embeddings generated by DNABERT. The study processes over 46 million high-quality DNA sequences to distinguish between cancerous and non-cancerous genomic material.

HantaBERT organization profile and shared community health files.