scRNA-seq for Microcephaly Research [IV]: Dirichlet Regression for Single-Cell Population Differences.

Area of Science:

• Computational Biology
• Bioinformatics
• Genomics

Background:

• Single-cell RNA sequencing (scRNA-seq) analysis involves cell clustering and population size determination.
• Comparing cellular composition across different experimental conditions is crucial for biological insights.
• Standard normalization methods create interdependent fractions, precluding direct statistical comparisons like t-tests.

Purpose of the Study:

• To introduce Dirichlet regression as a statistical method for analyzing changes in cellular composition from scRNA-seq data.
• To provide a robust framework for comparing multiple conditions with biological replicates.
• To offer a solution for the statistical challenges posed by normalized cell population data.

Main Methods:

• Utilized Dirichlet regression to model the proportions of cell clusters.
• Applied the method to scRNA-seq data with multiple experimental conditions and biological replicates (n>=3).
• Developed and demonstrated an example using R code for analysis and interpretation.

Main Results:

• Dirichlet regression effectively compares cellular composition shifts between conditions.
• The method accounts for the interdependence of normalized cell population fractions.
• The provided R code facilitates practical implementation and result interpretation.

Conclusions:

• Dirichlet regression is a powerful and appropriate statistical tool for scRNA-seq compositional data analysis.
• This method enables accurate comparison of cellularity changes across multiple experimental groups.
• The study offers a valuable computational resource for researchers in the field.