Constrained Pseudo-Time Ordering for Clinical Transcriptomics Data.

Area of Science:

• Genomics
• Computational Biology
• Translational Medicine

Background:

• Time series RNA sequencing (RNASeq) studies offer insights into disease progression and treatment responses.
• Integrating multi-patient RNASeq data is difficult due to patient heterogeneity and sparse time points, hindering accurate response pattern reconstruction.

Purpose of the Study:

• To develop a robust method for analyzing transcriptomics data in clinical and response studies.
• To accurately order samples along a biological response trajectory, accounting for individual patient variations.

Main Methods:

• A constrained-based pseudo-time ordering method was developed.
• Polynomials model gene expression dynamics over time.
• An Expectation-Maximization (EM) algorithm determines sample placement and model parameters.

Main Results:

• The method accurately assigns samples to their correct positions on a response curve.
• It successfully respects individual patient trajectories.
• Application to four datasets demonstrated improved ordering compared to previous methods.

Conclusions:

• The developed pseudo-time ordering method enhances the analysis of transcriptomics data in treatment response studies.
• It provides more accurate biological insights into disease dynamics and therapeutic effects.
• This approach addresses key challenges in integrating heterogeneous patient data.