psupertime: supervised pseudotime analysis for time-series single-cell RNA-seq data.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) enables high-resolution analysis of biological processes.
• Time-series scRNA-seq studies are increasingly common but lack specialized computational tools for inferring temporal gene expression dynamics.
• Existing unsupervised pseudotime methods are not optimized for identifying genes that vary coherently with time or for accurate temporal ordering.

Purpose of the Study:

• To introduce psupertime, a supervised computational approach for analyzing time-series scRNA-seq data.
• To accurately identify genes exhibiting coherent temporal expression patterns.
• To provide improved cell ordering and a classifier for new data with unknown sequential labels.

Main Methods:

• Developed psupertime, a supervised pseudotime method utilizing a regression model.
• Incorporated time-series labels as explicit input into the model.
• Evaluated performance against benchmark classifiers and unsupervised pseudotime techniques.

Main Results:

• psupertime effectively identifies genes that vary coherently along a time series.
• The method provides accurate pseudotime values for individual cells.
• psupertime outperforms benchmark classifiers in identifying time-varying genes and improves cell ordering compared to unsupervised methods.

Conclusions:

• psupertime is a fast and interpretable tool for analyzing scRNA-seq data with sequential labels (e.g., time, drug dosage, disease progression).
• The method facilitates the targeted identification of genes involved in specific biological processes over time.
• An R package is available for widespread use in the research community.