Evaluating thermodynamic models of enhancer activity on cellular resolution gene expression data

Area of Science:

• Developmental Biology
• Genomics
• Computational Biology

Background:

• High-throughput sequencing and transcriptomic technologies offer new ways to study gene regulation quantitatively.
• Current models of sequence-gene expression relationships have limitations.

Purpose of the Study:

• To address methodological challenges in quantitative gene expression modeling from sequence data.
• To develop and validate improved models using Drosophila embryo data.
• To identify novel regulatory factors influencing gene expression patterns.

Main Methods:

• Developed a framework for processing 3D gene expression data at cellular resolution.
• Proposed a novel scoring metric, weighted pattern generating potential (w-PGP), for model evaluation.
• Employed a systematic model-building strategy to manage complexity and prevent overfitting.

Main Results:

• Identified ZELDA, SLOPPY-PAIRED, and NUBBIN as transcription factors with significant regulatory influence.
• Demonstrated the superiority of w-PGP over existing scoring schemes.
• Showcased how small enhancer datasets can yield unreliable quantitative models.

Conclusions:

• The proposed w-PGP metric enhances the evaluation of gene expression models.
• Accurate quantitative modeling requires comprehensive datasets to avoid unreliable conclusions.
• New insights into the regulatory network of anterior-posterior patterning in Drosophila have been gained.