Machine learning discoveries of 2nd order combinations in peripheral blood mononuclear cells (PBMCs) strongly activated in fatal West African Ebola Virus (EBOV) Disease (EVD)

The pathogenesis of human Ebola virus disease (EVD) is complex and to clarify how host responses contribute to EVD pathophysiology, Amie J Eisfeld et. al. 2017 performed multi-platform 'omics analysis of peripheral blood mononuclear cells (PMBCs) and plasma from EVD patients. Their results pointed that EVD molecular signatures overlapped with those of sepsis, thus indicating that pancreatic enzymes contributed to tissue damage in fatal EVD, and suggested that Ebola virus infection might induce aberrant neutrophils (activities of which could explain hallmarks of fatal EVD). For this, they record Table S5 i.e file NIHMS922130-supplement-5.xlsx, in their supplementary, that contains the comparison for Fatalities vs. healthy controls (FvsH), to study the PMBC transcriptome data. This table contains a list of 11804 unique genes, for whom log2 fold change (logFC) and q-value (adj.P.Val), have been recorded. Due to limited computational power / resources, only genes that are (up)-regulated with a logFC >= 0.3, which accounts to a total number of 3327, and genes that are (down)-regulated with a logFC < -0.3, which accounts to a total number of 3529 genes, are considered. There are not much genomic studies which account for combinations of genes or factors that might be working synergistically in human EVD. Because, if that is done, it would open a range of insights into multiple factors that might be working synergistically at say, 2nd order level, to start with. These might help in developing therapeutics for the EVD, based on synergistic mechanisms. With this aim in mind, the idea of this current manuscript is to study the 2nd order rankings of genes in the FvsH case, using PMBC transcriptome data with respect to human EVD. A recently developed search engine ranked combinations of Y-X (X, a particular gene/protein) at 2nd order level for genes involved in EVD cases. These rankings reveal which Y-X combinations might be working synergistically in EVD. If found true, virologists can further test the combination of interest in wet lab and determine the mechanism of functioning between Y and X. In this research work, we cover various combinations of Y-X. Further, in my limited grasp, if proven true, these combinations might pave way for development of gene based therapies aimed at resolving pathological issues related to EVD.