psupertime

psupertime analyzes time-series single-cell RNA sequencing (scRNA-seq) data using a supervised pseudotime approach to identify temporal changes in gene expression and order cells according to experimental sequential labels.

Key Features:

• Supervised Pseudotime Analysis: Incorporates time-series sequential labels directly into the analysis to identify temporal transcriptome changes.
• Regression Model-Based Approach: Uses sequential labels as input to fit a regression model that focuses on genes varying coherently with those labels.
• Identification of Time-Varying Genes: Identifies genes exhibiting coherent temporal variation across the sequence of labels.
• Pseudotime Values and Cell Ordering: Computes pseudotime values for cells based on experimental sequential labels to provide an ordering aligned with conditions.
• Classifier for New Data: Applies the learned model to estimate labels for new datasets with unknown or differing labels, including contexts such as drug dosage and disease progression.

Scientific Applications:

• Sequentially labeled scRNA-seq datasets: Analysis of single-cell RNA-seq data that include time-series or other ordered experimental labels.
• Developmental biology: Studying developmental trajectories by identifying genes and ordering cells along developmental sequences.
• Disease progression: Investigating transcriptional changes associated with stages of disease progression.
• Treatment response and drug dosage studies: Characterizing gene expression dynamics and label estimation across treatment conditions or dosage gradients.

Methodology:

Uses a supervised regression model that integrates sequential time-series labels to model gene expression changes, compute pseudotime values, identify coherently varying genes, and classify labels for new data.