Abstract
Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disorder that leads to motor neuron degeneration, muscle weakness, and respiratory failure. Despite ongoing research, effective treatments for ALS are limited. This study aimed to apply network biology and machine learning (ML) techniques to identify novel repurposed drug candidates for ALS. In this study, we conducted a meta-analysis using 4 transcriptome data in ALS patients (including motor neuron and muscle tissue) and healthy controls. Through this analysis, we uncovered common shared differentially expressed genes (DEGs) separately for motor neurons and muscle tissue. Using common DEGs as proxies, we identified two distinct clusters of highly clustered differential co-expressed cluster genes: the 'Muscle Tissue Cluster' for muscle tissue and the 'Motor Neuron Cluster' for motor neurons. We then evaluated the performance of the nodes of these two modules to distinguish between diseased and healthy states with ML algorithms: KNN, SVM, and Random Forest. Furthermore, we performed drug repurposing analysis and text-mining analyses, employing the nodes of clusters as drug targets to identify novel drug candidates for ALS. The potential impact of the drug candidates on the expression of cluster genes was predicted using linear regression, SVR, Random Forest, Gradient Boosting, and neural network algorithms. As a result, we identified five novel drug candidates for the treatment of ALS: Nilotinib, Trovafloxacin, Apratoxin A, Carboplatin, and Clinafloxacin. These findings highlight the potential of drug repurposing in ALS treatment and suggest that further validation through experimental studies could lead to new therapeutic avenues.