Musaddiq K Lodi
Single-cell RNA sequencing (scRNA-seq) has transformed the study of cellular heterogeneity, enabling detailed characterization of cell types, regulatory programs, and disease-associated transcriptional changes. However, the growing number of computational methods for core analysis tasks introduces variability and limits reproducibility across datasets. In this dissertation, I present a generalizable computational framework for single-cell analysis that integrates regulatory pathways and cellular signaling to improve biological interpretability. First, I develop consensus-based methods for key scRNA-seq tasks, including cell clustering and gene regulatory network inference, leveraging a wisdom-of-crowds strategy to enhance performance across datasets. Second, I introduce novel approaches for cell-type-resolved marker gene selection and signaling-informed transcription factor activity estimation, incorporating both downstream gene expression and upstream cell–cell communication signals to better capture regulatory mechanisms. Finally, I apply state of the art single cell genomics analysis methods to disease contexts in neurology and cancer, using single-cell and spatial transcriptomics to uncover cell-type-specific signaling and transcriptional changes associated with disease progression and therapeutic response. Together, this work bridges methodological development and biomedical application, providing a unified framework for integrating cellular signaling and gene regulation in single-cell genomics.