Mohammad Naimul Islam Shanto

Computational Biology & Bioinformatics

Developing quantitative and machine-learning methods for biological data, with emphasis on interpretable phenotype discovery, single-cell heterogeneity, and computational analysis of high-dimensional biomedical measurements.

Live-Cell Dynamics & Trajectory Analysis

Studying how cell morphology and motility change over time using frame-level measurements, frame-to-frame transitions, trajectory windows, state occupancy, and behavioral-state transition analysis from live-cell time-lapse microscopy.

Machine Learning & Deep Representation Learning

Designing supervised, unsupervised, and self-supervised models for biomedical data, including temporal encoders, latent representations, neural-network architectures, classification, feature selection, and interpretable model comparison.

Dynamic Phenotyping & Unsupervised Discovery

Using methods such as UMAP, PHATE, Leiden clustering, Gaussian mixture models, and trajectory-state analysis to identify reproducible biological structure while carefully separating visualization, clustering, feature redundancy, and statistical inference.

Current Research: Dynamic Phenotyping of Breast Cell Trajectories

My current project compares MCF10A and MDA-MB-231 live-cell trajectories to investigate candidate morphology and motility signatures associated with metastatic behavior. The work integrates cell-level feature selection with frame-to-frame transition features, short-term trajectory representations, dimensionality reduction, clustering, and behavioral-state transition analysis.

A central methodological question is whether temporally ordered state-and-change representations reveal biologically meaningful cellular behaviors that are not visible from static snapshots or whole-cell summary statistics alone.