Prioritization and validation of asthma and respiratory allergy biomarkers using in silico analysis: Systems Biology

Abstract

Introduction. Asthma and respiratory allergy are chronic inflammatory diseases, with high prevalence and a wide clinical spectrum. Due to their heterogeneity, it is difficult to diagnose some patients and predict their response to treatments. Moreover, although allergic mechanisms have been implicated in most asthma diagnostics, there are still a 10-33% of patients with nonallergic asthma, less studied and understood. Thus, there is a need to define new biomarkers capable of classify patients correctly. At this respect, we defined a group of 94 potential biomarkers with the ability to differentiate clinical phenotypes and disease severity. Here, the objective was to theoretically prioritize those biomarkers using systems biology, based on their association with the studied diseases. Methods. Anaxomics’ TPMS technology (Therapeutic Performance Mapping System) was used to create one mathematical model, according to molecular motifs, per disease: respiratory allergy (RA), allergic asthma (AA) and nonallergic asthma (NA). The relationship of each candidate with the diseases was analyzed by artificial neural networks (ANNs) scores, according to their specificity. A validation of the theoretical results was performed through a study of their sensitivity and specificity, through ROC curve analysis, using gene expression data obtained from peripheral samples from healthy control subjects, RA patients, and asthmatic patients (AA and NA). Finally, a triggering analysis was performed, and possible pathways connecting triggering and specific proteins were created using Cytoscape program (Pathlinkers). Results. First, two molecular motifs were defined for RA, shared with AA; three motifs were specific for AA; and two for NA. According to these molecular motifs, 21 from the 94 candidate biomarkers showed the highest specificity for at least one of the diseases studied: 7 for RA, 12 for AA and 2 for NA. Regarding the experimental validation, ROC curves analysis highlighted 13 genes with the potential to discriminate between phenotypes and severity according to the AUC (Area Under the Curve) obtained, confirming the correlation between theoretical and experimental specificity results in some of the genes analyzed. Finally, this study also revealed the ability of AKT1, STAT1 and MAPK13 to trigger the three conditions, along with TLR4 in asthma; and possible pathways connecting the 4 triggering proteins with the 21 specific proteins were built. Conclusion. We theoretically prioritize 21 possible new molecular biomarkers according to their specificity with different respiratory diseases, and validated some of them. Also, 4 proteins were defined as triggers of the diseases, giving potential targets to future therapies.