Trajectory Inference with Cell-Cell Interactions (TICCI): intercellular communication improves the accuracy of...

Area of Science:

• Computational Biology
• Genomics
• Developmental Biology

Background:

• Single-cell transcriptome analysis is crucial for understanding cell differentiation and development.
• Existing trajectory inference algorithms struggle with high dimensionality, noise, and data-specific biological knowledge requirements.
• Intercellular communication plays a vital role in developmental processes.

Purpose of the Study:

• To introduce Trajectory Inference with Cell-Cell Interaction (TICCI), a novel method for single-cell transcriptome analysis.
• To address challenges in trajectory inference by integrating intercellular communication information at single-cell resolution.
• To leverage gene expression similarity patterns as a proxy for biomolecular information exchange between cells.

Main Methods:

• Constructing a cell-neighborhood matrix weighted by intercellular similarity and cell-cell interaction (CCI) information.
• Utilizing Louvain partitioning for trajectory branch identification and noise attenuation.
• Employing single-cell entropy (scEntropy) for differentiation status assessment.
• Applying the Chu-Liu algorithm for directed least-square modeling of trajectory branches.
• Implementing an improved diffusion fitted time algorithm for cell-fitted time computation in non-connected topologies.

Main Results:

• TICCI accurately reconstructs cell differentiation trajectories on single-cell RNA sequencing (scRNA-seq) datasets, correlating with known genealogical branching and gene markers.
• Verification using extrinsic labels confirms the utility of CCI information in improving trajectory inference accuracy.
• Comparative analysis demonstrates TICCI's superior performance in precise temporal ordering of cells.
• The method effectively attenuates noise and assesses differentiation status.

Conclusions:

• TICCI offers a robust and accurate approach to inferring cell differentiation trajectories by incorporating intercellular communication.
• The integration of CCI information enhances the precision of temporal ordering and trajectory reconstruction.
• TICCI provides a valuable tool for single-cell transcriptome analysis, overcoming limitations of previous methods.