Sazan Mahbub

BiRNA-BERT is a Transformer-based model pretrained on RNA sequences using adaptive tokenization, achieving state-of-the-art results in long-sequence tasks and efficient performance on short sequences.

This paper introduces an uncertainty-aware discrete diffusion model for protein design, improving sequence generation from 3D structures by incorporating position-specific uncertainty and pretrained models.

The paper introduces contextualized network inference to estimate sample-specific transcriptional networks for 7000 tumors, improving accuracy and enabling novel prognostic subtyping.

This paper introduces AIDO.DNA, a DNA language model that improves representation of DNA sequences across multiple tasks by scaling model size and training data.

AIDO.Protein is a mixture-of-experts model for protein understanding and design, achieving state-of-the-art performance with improved efficiency.

This paper introduces AIDO.RNA, a large-scale foundation model for RNA function and structure prediction, achieving state-of-the-art performance on multiple tasks.

Pair-EGRET predicts protein-protein interaction sites using graph attention networks and protein language models, outperforming existing methods on experimental data.

SAINT-Angle improves protein backbone torsion angle prediction using a self-attention network and transfer learning, outperforming existing methods on benchmark datasets.

QT-GILD improves species tree estimation by imputing missing quartets from incomplete gene trees using deep learning, enhancing accuracy despite data gaps.

EGRET improves PPI site prediction using a graph attention network and transfer learning from pre-trained models.

SAINT improves Q8 protein secondary structure prediction using self-attention and Deep3I network, outperforming existing methods on benchmark datasets.