A team of researchers from the University of Washington has introduced a new way to better understand how cells change over time. When scientists study individual cells using RNA sequencing, they often want to track how these cells develop or respond to treatments. This process, called pseudotime analysis, helps reconstruct the timeline of changes happening inside cells. However, current tools often struggle to make accurate predictions, especially when working with different types of biological data like DNA accessibility (scATAC-seq) or cell images.

To overcome these challenges, the researchers created Sceptic, a computational method based on support vector machines, a type of machine learning model. Sceptic is designed to be more accurate and more versatile than existing tools. It not only works well with RNA sequencing data but also performs reliably with other single-cell data types, including image-based data from cell nuclei. By accurately predicting the order in which cells progress through different states, Sceptic can help scientists understand complex biological processes such as cell differentiation, disease progression, or how cells respond to new drugs.

Sceptic analysis for time-series single-cell sequencing and imaging data

A Sceptic from the user’s perspective. Sceptic can take various types of single-cell or single-nucleus data as input and is trained to predict each cell’s time label (shown as a confusion matrix) and to infer the pseudotime for each cell (shown as a density plot across all cells). B Overview of Sceptic analysis

Sceptic’s strength lies in its ability to generalize across different datasets while maintaining high accuracy. The team tested it on multiple types of single-cell datasets and found that it consistently outperformed other popular methods. Because of this, Sceptic holds promise for researchers studying time-related changes in cells and could become a valuable tool in drug development, cancer research, and regenerative medicine.

Availability – The MIT-licensed Sceptic source code is available at GitHub (https://github.com/Noble-Lab/Sceptic) and Zenodo (https://doi.org/10.5281/zenodo.15729886)