Skip to main content


Fig. 1 | BMC Medical Genomics

Fig. 1

From: Novel disease syndromes unveiled by integrative multiscale network analysis of diseases sharing molecular effectors and comorbidities

Fig. 1

Overview of the study. a Preprocessing. We created a controlled disease terminology across the molecular data from the GWAS Catalog and the HCUP clinical datasets (Methods- Data preprocessing to define disease bundles). We mapped GWAS diseases into disease bundles, i.e., group diseases, using the EMBL-EBI EFO, UMLS-CUI, SNOMED-CT nomenclatures integrated with expert curation (Methods- Creation of the SNOMED-coded disease-bundles from GWAS terms). Similarly, we mapped HCUP diseases coded with ICD-9-CM terminology into disease bundles by using SNOMED-CT, UMLS, and expert curation (Methods- Mapping HCUP diseases to disease-bundles). b eQTL RNA overlap model. Convergence between downstream eQTLs signals associated with coding and intergenic disease-associated polymorphisms are calculated for each pair of diseases (Methods- Statistical overlap). We selected significant disease pairs sharing convergent mechanisms by applying the Fisher’s Exact Test (FET) according to the contingency table shown in the panel. We considered significant disease pairs surpassing FDReRNA of 0.05. c Disease comorbidity model. We computed the disease comorbidity for each disease pairs by applying logistic regression (Methods- Calculation of disease comorbidity) to the clinical datasets. The effect size and significance of disease co-occurrence in clinical datasets (comorbidities) were controlled for age, gender, and race. Significant comorbid disease pairs were selected accordingly with FDR values (FDRcomorbidity < 0.05). d Comparative study. Finally, congruence between molecular-prioritized disease pairs and clinically-prioritized comorbidities is measured by applying FET-based enrichment studies (FETfinal) (Methods- Comparative studies between eQTLs and HCUP). e Network visualization. We further investigated in detail the molecular networks of comorbid disease pairs with sharing convergent genetics (eQTL RNAs) (Methods- Network visualization of the comorbidities sharing intergenic genetic risks). f Curation. For additional validation, we conducted a systematic curation of the literature using PUBMED and Google Scholar for the comorbidities discovered from HCUP datasets (FDR < 0.05, OR > 3) having convergent eQTL RNAs (Methods- Curation of prioritized comorbidities)

Back to article page