Automated gene oscillation phase classification for zebrafish presomitic mesoderm cells

Area of Science:

• Developmental biology
• Cellular and molecular biology
• Genetics and genomics

Background:

• Zebrafish somitogenesis relies on a segmentation clock controlling gene expression oscillations in presomitic mesoderm (PSM) cells.
• These oscillations, divided into eight phases by mRNA co-localization, are crucial for developmental research but challenging to analyze manually.
• Accurate identification of gene oscillation phases is vital for understanding zebrafish development.

Purpose of the Study:

• To establish an effective automated framework for classifying gene oscillation phases in zebrafish PSM cell images.
• To develop a computational method that overcomes the time-consuming and difficult nature of manual analysis.
• To improve the efficiency and accuracy of analyzing gene expression patterns during zebrafish somitogenesis.

Main Methods:

• A two-stage segmentation procedure to identify individual zebrafish PSM cells.
• Extraction of multiple features from cell patches to quantify subcellular mRNA distribution.
• Application of a support vector machine (SVM) with a combined kernel for feature fusion and classification.

Main Results:

• The developed framework achieved reasonably good performance in classifying gene oscillation phases.
• Novel feature sets, NF9 and SPIN, demonstrated superior performance compared to other cell features.
• Kernel fusion effectively combined heterogeneous features (numerical and histogram-based), outperforming single-feature classification.

Conclusions:

• The automated framework provides an effective solution for classifying gene oscillation phases in zebrafish PSM cells.
• The proposed feature sets and kernel fusion method enhance classification accuracy for this biological problem.
• This approach facilitates more efficient and reliable research into zebrafish segmentation clock mechanisms.