iProm-Zea

iProm-Zea predicts and classifies plant promoters, distinguishing TATA box-containing and TATA-less promoters to support studies of transcriptional regulation.

Key Features:

• Two-Layer Model Architecture: A dual-layer framework where the first layer discriminates promoters from non-promoters and the second layer classifies identified promoters as TATA or TATA-less.
• Convolutional Neural Network (CNN) Utilization: Employs CNNs within the model to capture sequence patterns for promoter identification and classification.
• Feature Encoding Scheme: One-hot encoding was selected as the optimal feature representation after comparative testing of encoding methods and machine learning algorithms.
• Validation and Performance: Performance was assessed using 5-fold cross-validation and evaluated through cross-species analyses across plant genomes.

Scientific Applications:

• Transcriptional regulation analysis: Provides annotations of promoter presence and type to inform studies of promoter function in transcriptional regulation.
• TATA-related post-transcriptional studies: Distinguishes TATA and TATA-less promoters to support research on the influence of the TATA box on downstream processes.
• Cross-species promoter prediction: Enables comparative promoter analysis across different plant species as demonstrated by cross-species evaluation.

Methodology:

The model uses a two-layer architecture implemented with convolutional neural networks; various machine learning algorithms were tested, one-hot encoding was selected after evaluation, performance was validated by 5-fold cross-validation, and cross-species analyses were performed.