Performance Comparison Between Deep Neural Network and Machine Learning Based Classifiers for Huntington Disease...

Area of Science:

• Genetics
• Computational Biology
• Machine Learning

Background:

• Huntington Disease (HD) is a neurodegenerative disorder impacting psychiatric, motor, and sleep functions.
• Early detection of HD is crucial for timely intervention and management.
• Deep Learning (DL) offers potential for developing advanced diagnostic tools.

Purpose of the Study:

• To develop and evaluate a Deep Neural Network (DNN) model for predicting Huntington Disease (HD) from human DNA sequences.
• To assess the efficacy of DNN and other machine learning classifiers in identifying HD genetic markers.

Main Methods:

• Human DNA sequences were obtained from NCBI and supplemented with synthetic data.
• DNA sequences were numerically converted using Chaos Game Representation (CGR).
• Features were extracted, including statistical measures and nucleotide counts (adenine, thymine, guanine, cytosine).
• Extracted features were used to train DNN, Neural Network (NN), Support Vector Machine (SVM), Random Forest (RF), and Classification Tree with Forward Pruning (CTWFP) models.

Main Results:

• DNN, NN, SVM, and RF models achieved 100% accuracy in predicting HD.
• The Classification Tree with Forward Pruning (CTWFP) model demonstrated 87% accuracy.
• Performance was evaluated using Accuracy, Sensitivity, Specificity, Precision, F1 score, and MCC.

Conclusions:

• Deep Neural Network (DNN) models show high potential for accurate and early detection of Huntington Disease (HD) through DNA sequence analysis.
• Machine learning approaches, particularly DNN, NN, SVM, and RF, offer robust solutions for identifying genetic predispositions to HD.
• The study highlights the effectiveness of CGR and feature extraction in preparing DNA data for machine learning-based disease prediction.