Clustering methods for analyzing large data sets: gonad development, a study case

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Data set repositories enable collation of experimental results from diverse sources.
• Heterogeneous data from multiple experiments presents challenges in quality assessment and analysis.
• Gene clustering on large datasets requires efficient and robust methodologies.

Purpose of the Study:

• To present methods for assessing the quality of heterogeneous biological data.
• To introduce a rapid gene clustering method using symbolic representation of variation profiles.
• To demonstrate the construction of transcriptional networks for enhanced gene clustering.

Main Methods:

• Data quality assessment focusing on reproducibility and similarity of experimental points.
• Translating variation profiles into symbolic chains for rapid gene clustering.
• Utilizing gene functional descriptions to identify informative comparisons for clustering.
• Extending the symbolic approach to construct transcriptional networks.

Main Results:

• Established routine methods for evaluating the quality of collated experimental data.
• Developed a rapid gene clustering method based on symbolic representation of variation profiles.
• Demonstrated that functional gene descriptions can guide the selection of informative comparisons.
• Showcased the construction of transcriptional networks from variation profiles for gene clustering.

Conclusions:

• The symbolic approach provides an effective strategy for clustering genes from large, heterogeneous datasets.
• Transcriptional networks derived from this method serve as powerful tools for gene clustering and biological network inference.
• The presented methods enhance the analysis of complex biological data from multiple sources.