C-ZIPTF: stable tensor factorization for zero-inflated multi-dimensional genomics data
Assistant Professor of Applied Mathematics
University of Massachusetts, Boston
Abstract: In this talk, we will delve into the challenges associated with tensor decompositions and present innovative models designed to tackle these obstacles effectively. We will present our innovative approach, Consensus-Zero Inflated Poisson Tensor Factorization (C-ZIPTF), which provides a stable tensor factorization method tailored for zero-inflated multi-dimensional data—a common phenomenon in genomics.
Our proposed method will be evaluated on synthetic zero-inflated count datasets, simulated single-cell RNA sequencing (scRNA-seq) data, and real-world multi-sample multi-condition scRNA-seq datasets. The results demonstrate that C-ZIPTF consistently uncovers known and biologically meaningful gene expression programs in both synthetic and real scRNA-seq data.
This presentation will underscore the high potential of tensor methods in advancing the analysis of complex genomic data, highlighting their implications for biological discovery and research methodologies in genomics. Join us to explore how C-ZIPTF can enhance our understanding of gene expression dynamics and contribute to the identification of critical biological patterns, paving the way for new insights in the field.