A hybrid model for the prediction of mRNA polyadenylation signals

Area of Science:

• Molecular Biology
• Bioinformatics
• Genomics

Background:

• mRNA polyadenylation is essential for cellular function, involving the addition of adenosine tails to mature mRNAs.
• Dysfunctional polyadenylation is linked to various human diseases, highlighting the need for accurate prediction methods.

Purpose of the Study:

• To develop a novel feature extraction approach for improved mRNA polyadenylation signal prediction.
• To evaluate the efficacy of a real-coded genetic neural network model trained on these extracted features.

Main Methods:

• Utilized K-gram nucleotide acid patterns, Position Weight Matrix (PWM), and Increment of Diversity (ID) for feature representation.
• Applied Principle Component Analysis (PCA) for dimensionality reduction and feature transformation.
• Trained a real-coded genetic neural network (GA-BP) model on the processed features.

Main Results:

• The proposed GA-BP algorithm achieved high performance metrics on a specific dataset.
• Achieved an accuracy of 82.98%, specificity of 82.95%, and sensitivity of 83.01%.

Conclusions:

• The GA-BP algorithm demonstrates significant potential as a tool for predicting mRNA polyadenylation signals.
• This approach offers a promising direction for understanding and diagnosing diseases related to polyadenylation malfunction.