Mouse blood cells types and aging prediction using penalized Latent Dirichlet Allocation

Area of Science:

• Computational biology
• Genomics
• Aging research

Background:

• Aging is a multifaceted process influenced by genomic, epigenomic, and transcriptomic changes.
• Single-cell RNA sequencing (scRNA-seq) offers high-resolution gene expression data for individual cells, revolutionizing aging studies.
• Novel computational methods are essential for extracting meaningful insights from complex scRNA-seq data.

Purpose of the Study:

• To introduce and apply a novel statistical method, penalized Latent Dirichlet Allocation (pLDA), to aging research.
• To develop a computational pipeline for analyzing mouse blood scRNA-seq data.
• To predict cell types and organismal aging status using gene expression profiles.

Main Methods:

• Application of penalized Latent Dirichlet Allocation (pLDA), a statistical method, to scRNA-seq data.
• Development of a computational pipeline that preprocesses scRNA-seq expression counts using pLDA.
• Prediction of cell types and aging status from the dimension-reduced data representation.

Main Results:

• The pLDA method was successfully applied to an aging mouse blood scRNA-seq dataset.
• A functional pipeline was established for cell type and aging prediction.
• The method demonstrated the ability to identify cell types and predict the age of individual cells.

Conclusions:

• pLDA effectively learns a reduced-dimension representation of gene expression profiles.
• This representation facilitates the identification of distinct cell types.
• The learned representation possesses predictive power for cellular age, contributing to aging mechanism understanding.