A machine learning and directed network optimization approach to uncover TP53 regulatory patterns.

Area of Science:

• Genomics
• Cancer Biology
• Bioinformatics

Background:

• TP53 is the most frequently mutated gene in human cancers.
• Understanding TP53 regulation is crucial for cancer research and treatment.

Purpose of the Study:

• To predict TP53 mutation status from transcriptomic data using machine learning.
• To reconstruct the regulatory effects of TP53 mutations on its target genes using directed networks.

Main Methods:

• Utilized data from Cancer Cell Line Encyclopedia and The Cancer Genome Atlas.
• Applied machine learning for mutation status prediction.
• Employed directed network optimization to infer TP53 regulatory profiles.

Main Results:

• Machine learning predicted TP53 mutation status but could not differentiate specific mutations.
• Directed network optimization successfully inferred TP53 regulatory profiles across various mutations and stress conditions (irradiation, hypoxia).
• Observed differential regulatory profiles influenced by mutation type, severity, hotspots, protein changes, and stress.

Conclusions:

• Directed regulatory networks offer a powerful approach to characterize the functional consequences of TP53 mutations.
• This method can be extended to other genetic perturbations, aiding drug design and precision medicine strategies.