DeepCBA: A deep learning framework for gene expression prediction in maize based on DNA sequences and chromatin...

Area of Science:

• Genomics
• Computational Biology
• Plant Science

Background:

• Chromatin interactions influence gene expression and traits by connecting regulatory elements to target genes.
• Current gene expression prediction methods often overlook chromatin interactions, limiting accuracy and regulatory element discovery.
• Maize (Zea mays) gene regulation is complex, necessitating advanced computational tools for accurate expression prediction.

Purpose of the Study:

• To develop a highly accurate deep learning model, DeepCBA, for predicting gene expression in maize by incorporating chromatin interaction data.
• To evaluate DeepCBA's performance against existing methods for gene expression classification and value prediction.
• To identify novel regulatory motifs and elements influencing gene expression and validate the model's utility in gene function exploration and breeding.

Main Methods:

• Development of DeepCBA, a deep learning model utilizing maize chromatin interaction data.
• Comparative analysis of DeepCBA against traditional methods using Pearson correlation coefficients (PCCs).
• Identification and characterization of important motifs enriched in specific genomic regions and exhibiting tissue specificity.

Main Results:

• DeepCBA achieved high accuracy in gene expression prediction, with PCCs reaching 0.929 when considering both proximal and distal interactions.
• The model significantly outperformed traditional methods, showing substantial increases in PCCs across different interaction types.
• DeepCBA identified biologically relevant motifs, validated through experimental analysis of maize genes (ZmRap2.7, ZmTb1) and promoter editing (ZmCLE7, ZmVTE4).

Conclusions:

• DeepCBA provides a powerful and accurate approach for predicting gene expression in maize by leveraging chromatin interaction data.
• The model facilitates the discovery of functional regulatory elements and offers insights into gene regulation.
• DeepCBA demonstrates significant potential for precise gene expression design and applications in intelligent breeding strategies.