Joint representation and visualization of derailed cell states with Decipher

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Comparing biological conditions (e.g., disease vs. health) is crucial for insights.
• Existing computational tools struggle to integrate single-cell genomics data across conditions.
• Characterizing transitions from normal to deviant cell states remains a challenge.

Purpose of the Study:

• To present Decipher, a novel deep generative model for characterizing derailed cell-state trajectories.
• To enable joint modeling and visualization of gene expression and cell states from single-cell RNA-seq data.
• To identify shared and disrupted cellular dynamics across normal and perturbed conditions.

Main Methods:

• Development of Decipher, a deep generative model.
• Joint modeling of gene expression and cell states from single-cell RNA-sequencing (scRNA-seq) data.
• Application to diverse datasets including pancreatitis, acute myeloid leukemia, and gastric cancer.

Main Results:

• Decipher effectively models and visualizes cellular dynamics across different conditions.
• The model reveals shared and disrupted gene expression patterns during cell state transitions.
• Superior performance demonstrated across multiple disease contexts.

Conclusions:

• Decipher provides a powerful computational tool for integrating and analyzing single-cell genomics data.
• The model facilitates the characterization of aberrant cell states and disease progression.
• Decipher advances our understanding of cellular dynamics in various pathological conditions.