An information theoretic treatment of sequence-to-expression modeling

Area of Science:

• Computational Biology
• Systems Biology
• Genetics

Background:

• Studying gene regulatory mechanisms traditionally relies on intuition-driven experiments.
• Existing methods for identifying gene regulators, enhancers, and transcription factor influence are often tedious and lack systematic strategies.

Purpose of the Study:

• To develop a systematic, information theory-based approach for designing gene regulation experiments.
• To enable rigorous experiment design strategies by quantifying the value of potential experiments.

Main Methods:

• Utilized a state-of-the-art mathematical model of gene expression to formalize knowledge of cis- and trans- regulatory mechanisms.
• Built an ensemble of plausible models representing uncertainties in gene regulatory knowledge.
• Quantified experiment value as information gain (reduction in entropy) by analyzing changes in the model ensemble and probability distribution.

Main Results:

• Developed a novel, formalized information-theoretic framework for reasoning about gene regulation experiments.
• Evaluated the information gain of various perturbation experiments on two developmental genes in D. melanogaster.
• Provided objective, biologist-friendly descriptions of information gained from experiments.

Conclusions:

• The presented information-theoretic approaches can guide systematic strategies for designing gene regulation studies.
• This framework offers a rigorous method to assess and maximize the value of experimental interventions in understanding gene regulation.