ZIAQ

ZIAQ applies zero-inflation-adjusted quantile regression to detect differential expression in single-cell RNA sequencing (scRNA-seq) data, explicitly addressing dropout events and multimodal expression distributions.

Key Features:

• Zero-Inflation Adjustment: Implements zero-inflation-adjusted quantile regression to model dropout events where expressed genes exhibit zero or low read counts.
• Quantile Regression Framework: Uses quantile regression to model multimodal and complex expression distributions typical of scRNA-seq datasets.
• Integrated Modeling Approach: Jointly accounts for dropout rates and complex data distributions within a unified model to improve differential expression detection.

Scientific Applications:

• Differential Expression Benchmarking: Demonstrated improved performance on simulated scRNA-seq datasets for identifying differentially expressed genes.
• Glioblastoma Cell-Type Analysis: Applied to human glioblastoma data comparing neoplastic versus non-neoplastic cells, improving ranking of biologically relevant genes and identification of disease-related pathways.

Methodology:

Zero-inflation-adjusted quantile regression and quantile regression modeling of multimodal expression distributions are combined in a unified model; the implementation is in R.