Bishal Shrestha

310K Ungar Building

University of Miami

1365 Memorial Drive

Coral Gables, FL 33124

I’m Bishal Shrestha, a Doctoral Researcher in Computer Science at the University of Miami, where my research bridges artificial intelligence and biology—developing deep learning models to explore the complex architecture of the genome and its role in human health and disease. My work focuses on building computational tools that help us understand how DNA folds and functions across space and time, especially through the lens of single-cell genomics and 3D chromatin organization.

Originally from Kathmandu, Nepal, I began my academic journey in computer engineering, driven by curiosity and a passion for innovation. Over the years, that curiosity evolved into a deep commitment to advancing biological discovery through computation. From building neural networks for genomic data to designing systems that detect changes in genome structure during development or disease, I’m excited by the challenge of turning data into insight.

Beyond research, I’m passionate about scientific communication, mentoring, and contributing to the global conversation around AI in biology. I believe that impactful science comes not just from good models, but from meaningful collaborations, clear storytelling, and a relentless drive to understand the unknown.

news

Apr 15, 2026	Protein–RNA quality assessment paper in Bioinformatics
Apr 01, 2026	SCW published in BMC Bioinformatics
Mar 15, 2026	ARC submitted to PROTEINS for CASP16
Feb 15, 2026	New paper in EMBO Molecular Medicine on CITED2
Feb 01, 2026	Serving as a reviewer for ISMB 2026 (Intelligent Systems for Molecular Biology), the flagship conference of the International Society for Computational Biology.

latest posts

Aug 26, 2020	Quantum Teleportation Simplified
Aug 23, 2020	Backstory on Hacking
Aug 23, 2020	How I became Executive Member of an Undergrad Club

selected publications

CITED2 is a druggable epigenetic switch coupling neuronal maturation to regenerative decline

F. Müller, E. McLachlan, A. Costa, and 14 more authors

EMBO Molecular Medicine, 2026
Inferring the qualities of protein-RNA models with graph transformers

Andrew Jordan Siciliano, Yiheng Bao, Bishal Shrestha, and 1 more author

Bioinformatics, 2026
ARC: Assessment of Interface Residue Conformation using an Ensemble of Graph Neural Networks

Bishal Shrestha, Andrew Jordan Siciliano, Gabriel Huang, and 2 more authors

2026

Manuscript ID 5704351, submitted to PROTEINS: Structure, Function, and Bioinformatics (Wiley), Mar 2026
HiC4D-SPOT: a spatiotemporal outlier detection tool for Hi-C data

Bishal Shrestha, and Zheng Wang

Briefings in bioinformatics, Jul 2025

Abs DOI HTML PDF

The 3D organization of chromatin is essential for the functioning of cellular processes, including transcriptional regulation, genome integrity, chromatin accessibility, and higher order nuclear architecture. However, detecting anomalous chromatin interactions in spatiotemporal Hi-C data remains a significant challenge. We present HiC4D-SPOT, an unsupervised deep-learning framework that models chromatin dynamics using a ConvLSTM-based autoencoder to identify structural anomalies. Benchmarking results demonstrate high reconstruction fidelity, with Pearson Correlation Coefficient and Spearman Correlation Coefficient values of 0.9, while accurately detecting deviations linked to temporal inconsistencies, topologically associating domain (TAD) and loop perturbations, and significant chromatin remodeling events. HiC4D-SPOT successfully identifies swapped time points in a time-swap experiment, captures simulated TAD and loop disruptions with high confidence scores and statistical significance of 0.01, and detects HERV-H boundary weakening during cardiomyocyte differentiation, as well as cohesin-mediated loop loss and recovery—aligning with experimentally observed chromatin remodeling events. These findings establish HiC4D-SPOT as an efficient tool for analyzing 3D chromatin dynamics, enabling the detection of biologically significant structural anomalies in spatiotemporal Hi-C data.
scHiGex: predicting single-cell gene expression based on single-cell Hi-C data

Bishal Shrestha, Andrew Jordan Siciliano, Hao Zhu, and 2 more authors

NAR Genomics and Bioinformatics, Jan 2025

Abs DOI HTML PDF

A novel biochemistry experiment named HiRES has been developed to capture both the chromosomal conformations and gene expression levels of individual single cells simultaneously. Nevertheless, when compared to the extensive volume of single-cell Hi-C data generated from individual cells, the number of datasets produced from this experiment remains limited in the scientific community. Hence, there is a requirement for a computational tool that can forecast the levels of gene expression in individual cells using single-cell Hi-C data from the same cells. We trained a graph transformer called scHiGex that accurately and effectively predicts gene expression levels based on single-cell Hi-C data. We conducted a benchmark of scHiGex that demonstrated notable performance on the predictions with an average absolute error of 0.07. Furthermore, the predicted levels of gene expression led to precise categorizations (adjusted Rand index score 1) of cells into distinct cell types, demonstrating that our model effectively captured the heterogeneity between individual cell types. scHiGex is freely available at https://github.com/zwang-bioinformatics/scHiGex.