Interspecies predictions of growth traits from quantitative transcriptome data acquired during fruit development

Area of Science:

• Systems biology
• Genomics
• Plant science

Background:

• Linking genotype to phenotype is a key biological challenge.
• Machine learning is increasingly vital in systems biology for understanding complex traits.
• Fruit development, including relative growth rate (RGR), is influenced by gene regulation, metabolism, and environmental factors.

Purpose of the Study:

• To predict fruit growth traits using multispecies transcriptomic data.
• To identify gene expression patterns associated with fruit development mechanisms.
• To evaluate the efficacy of generalized linear models (GLMs) in predicting phenotypic traits.

Main Methods:

• Multispecies transcriptomic analysis of nine fruit types.
• Comparative transcriptomic analysis using multivariate methods.
• Prediction of growth traits (RGR, developmental progress, fruit weight, protein content) using GLMs.
• Identification of key gene ontology (GO) terms and orthogroups.

Main Results:

• Transcriptome profiles showed similar patterns across species.
• Metabolic process genes, especially those related to cell wall carbohydrates and proteins, were most predictive of growth.
• Incorporating a time lag improved RGR prediction, highlighting the role of protein synthesis.
• GLMs effectively predicted growth traits based on multispecies transcriptomes.

Conclusions:

• Gene expression, particularly metabolic pathways, can accurately predict fruit growth traits.
• Multispecies transcriptomic data combined with machine learning offers a powerful approach to understanding plant development.
• Considering temporal dynamics, like protein production lag, enhances predictive accuracy for phenotypic traits.