Generalized and scalable trajectory inference in single-cell omics data with VIA.

Area of Science:

• Computational Biology
• Single-Cell Omics Data Analysis
• Bioinformatics

Background:

• Inferring cellular trajectories from omic data is crucial for understanding cell fate decisions.
• Challenges include large data size, data type diversity, and complex topological structures.
• Existing methods struggle with scalability and reconstructing non-tree-like pathways.

Purpose of the Study:

• To present VIA, a scalable algorithm for accurate cellular trajectory inference.
• To overcome limitations of existing methods in handling large and complex single-cell datasets.
• To enable discovery of elusive cell lineages and fates.

Main Methods:

• Utilizes lazy-teleporting random walks for trajectory reconstruction.
• Designed for scalability to handle millions of cells.
• Applicable to diverse omic data types (transcriptomic, proteomic, epigenomic, multi-omics, morphological).

Main Results:

• VIA accurately reconstructs complex cellular trajectories, including cyclic and disconnected structures.
• Successfully applied to a 1.3-million-cell mouse transcriptomic atlas, preserving global connectivity.
• Identified elusive lineages and rare cell fates missed by other methods.

Conclusions:

• VIA is a robust and efficient tool for single-cell trajectory inference.
• Enables deeper biological discovery from large-scale, complex omic datasets.
• Advances the field of single-cell data science by providing a scalable and versatile solution.