CTSV

CTSV detects cell-type-specific spatially variable genes by modeling sparse spatial transcriptomic count data with zero-inflation and overdispersion while incorporating cell-type proportions.

Key Features:

• Zero-Inflated Negative Binomial Regression: CTSV directly models sparse raw count data through a zero-inflated negative binomial regression model that accounts for excess zeros and overdispersion.
• Cell-Type Proportion Integration: CTSV incorporates cell-type proportions and spatial effect functions within the zero-inflated negative binomial regression framework.
• Hypothesis Testing with pscl: CTSV performs hypothesis testing based on the R package pscl to fit the model and output p-values and q-values for genes in each cell type.
• Cauchy Combination Rule: CTSV applies a Cauchy combination rule to integrate P-values from multiple choices of spatial effect functions for robustness.
• Scalability: CTSV scales to datasets with tens of thousands of genes measured on hundreds of spots.

Scientific Applications:

• Cell-Type-Specific Spatial Variable Gene Detection: CTSV identifies spatially variable genes at the cell-type level rather than only at aggregated tissue levels.
• Pancreatic Ductal Adenocarcinoma Analysis: CTSV reveals biological insights at the cell-type level by analyzing pancreatic ductal adenocarcinoma spatial transcriptomic data.

Methodology:

CTSV models spatial raw count data using a zero-inflated negative binomial distribution with cell-type proportions and spatial effect functions in a regression framework, fits the model using the R package pscl, and integrates P-values across multiple spatial effect function choices via a Cauchy combination rule.