System biology based integrative analysis to predict early biomarkers for COVID-related MIS in children

The outbreak of COVID-19 caused by SARS-CoV-2 has introduced unparalleled difficulties, including the appearance of a rare but serious condition in children referred to as Multisystem Inflammatory Syndrome in Children (MIS-C). This syndrome, marked by an overactive immune response resulting in extensive inflammation throughout various organ systems, has underscored the pressing necessity for predictive biomarkers to enhance early diagnosis and treatment. This research aimed to fill this gap using a comprehensive bioinformatics approach, applying fold-change and adjusted p-value techniques to detect differentially expressed genes (DEGs). DESeq2 was utilized to enhance these biomarkers, ensuring a strong and precise selection procedure. To accurately predict MIS-C biomarkers, numerous machine learning classifiers were evaluated, and the Random Forest model excelled compared to the rest. It showed a significant accuracy of 0.81 for moderate DEGs, confirming its appropriateness for biomarker prediction in COVID-19-related MIS-C. The results highlight the strength of computational biology in understanding the molecular processes associated with intricate conditions such as MIS-C, facilitating focused exploration of treatment approaches. Through in silico analysis and visualisation, a particular drug ligand named Methylprednisolone came out to an ideal candidate that shows favourable binding affinity with the genes deduced from ML analysis. This study not only clarifies the pathophysiology of MIS-C but also opens avenues for utilizing advanced statistical and machine learning techniques in clinical studies. By connecting computational and clinical fields, this study marks an important advancement in reducing the effects of severe COVID-19-related issues in children.