iProm-Zea: A two-layer model to identify plant promoters and their types using convolutional neural network

Area of Science:

• Genomics and Molecular Biology
• Bioinformatics and Computational Biology

Background:

• Promoters are crucial DNA sequences regulating gene transcription.
• Accurate promoter identification is vital for understanding gene regulation.
• Existing in silico tools often lack the ability to differentiate promoter types like TATA and TATA-less.

Purpose of the Study:

• To develop an accurate computational tool, iProm-Zea, for identifying DNA promoters.
• To classify promoters into TATA and TATA-less categories using a two-layer predictor.
• To optimize feature encoding and model selection for enhanced prediction accuracy.

Main Methods:

• Developed a two-layer predictor, iProm-Zea, utilizing convolutional neural network (CNN).
• Employed and evaluated four feature encoding schemes, selecting one-hot encoding for optimal performance.
• Utilized 5-fold cross-validation for rigorous testing and performance assessment.

Main Results:

• iProm-Zea demonstrated significant potential in accurately identifying promoters and classifying them as TATA or TATA-less.
• The predictor showed promising performance in cross-species analysis, indicating broad applicability.
• A user-friendly web server was established for easy access by researchers.

Conclusions:

• iProm-Zea provides a valuable in silico tool for promoter identification and classification.
• The developed CNN model and one-hot encoding scheme are effective for this task.
• The freely accessible web server facilitates advancements in transcriptional regulation research.