Machine Learning Inference of Gene Regulatory Networks in Developing Mimulus Seeds

Area of Science:

• Evolutionary Biology
• Plant Science
• Genetics

Background:

• Angiosperm seeds are key evolutionary innovations driving plant diversification.
• Postzygotic reproductive barriers, like hybrid seed inviability, are crucial for speciation.
• The Mimulus genus provides a model for studying seed-based reproductive isolation.

Purpose of the Study:

• To investigate gene regulatory network (GRN) inference in developing Mimulus hybrid seeds.
• To identify gene regulatory mechanisms critical for seed development and reproductive barriers.
• To generate a reference set of putative gene regulations in Mimulus.

Main Methods:

• Utilized time-series RNA-sequencing data from viable Mimulus hybrid seeds.
• Applied two GRN inference algorithms: RTP-STAR and KBoost.
• Analyzed transcriptomic data subsets to infer gene regulatory relationships.

Main Results:

• Successfully inferred GRNs from Mimulus hybrid seed transcriptomic data.
• Identified overlapping and distinct gene regulations between the two algorithms.
• Highlighted potential novel regulatory mechanisms involved in seed development.

Conclusions:

• GRN inference is a powerful approach for uncovering gene regulatory mechanisms in plant seeds.
• The study provides insights into the genetic basis of reproductive barriers in Mimulus.
• Identified putative gene regulations warrant further investigation for their role in speciation.