Amarda Shehu

This paper investigates whether protein structure improves mutation-induced stability prediction, finding that sequence embeddings dominate and structure provides limited additive value in most cases.

This study benchmarks fourteen protein-language models (pLMs) to determine which best captures biological properties, finding that mid-scale models often outperform larger ones for general tasks.

The paper develops a deep surrogate model to generate DNA breathing features efficiently, improving transcription factor binding predictions.

The paper introduces JAX-EPBD, a high-throughput framework for simulating DNA breathing dynamics, showing 30x speedup and improved predictions of transcription factor binding using SELEX datasets.

The paper improves TF binding site prediction by integrating DNA biophysical properties with sequence data using a multi-modal deep learning model, achieving better performance in both in vivo and in vitro datasets.

The paper introduces pyDNA-EPBD, a software tool for simulating DNA breathing dynamics using the EPBD model, and demonstrates its ability to calculate DNA bubble probabilities and mutation effects.

This paper introduces GOProFormer, a transformer-based method for Gene Ontology protein function prediction, showing superior performance over existing methods and proposing a new benchmark dataset.