Alzheimer-Compound Identification Based on Data Fusion and forgeNet_SVM

Area of Science:

• Computational chemistry
• Drug discovery
• Bioinformatics

Background:

• Alzheimer's disease (AD) drug development requires efficient screening of candidate compounds.
• Current methods face challenges in success rate, cost, and workload.

Purpose of the Study:

• To introduce forgeNet_SVM, a novel algorithm for identifying Alzheimer's-related compounds.
• To improve the accuracy and efficiency of compound screening for AD drug development.

Main Methods:

• Compounds were collected and their features extracted using ECFP6, MACCS, and RDKit descriptors.
• The forgeNet component identified important features from a fused feature set.
• Support Vector Machines (SVM) classified compounds using the selected features.

Main Results:

• The selected feature set outperformed the complete and individual feature sets.
• forgeNet_SVM demonstrated superior accuracy in identifying Alzheimer's-related compounds compared to classical classifiers.
• Performance metrics including TPR, FPR, Precision, Specificity, F1, and AUC confirmed the algorithm's effectiveness.

Conclusions:

• forgeNet_SVM offers a more accurate and efficient approach for identifying potential Alzheimer's drug candidates.
• The algorithm shows promise for accelerating drug discovery and development for Alzheimer's disease.
• Feature selection is crucial for enhancing the performance of compound identification algorithms.