Fig 1.
PRISMA flow chart of the transcriptional meta-analysis.
(A)Selection of eligible GEO datasets for systems biology analysis according to PRISMA 2019 flow diagram. (B) Flow diagram of bioinformatics analysis. (C) A list of packages and functions.
Fig 2.
Differentially expressed mRNA transcripts among individuals with different types of dysglycemia.
Volcano plots display the differentially expressed genes based on adjusted p-value and fold-difference variation of the gene expression in comparisons between each clinical group and the reference group (normoglycemic healthy controls) as follows: T1DM (A), T2DM (B) and pre-DM (C). In (D), Venn diagram shows genes with significant adjusted p-values (<0.05) of each comparison between the indicated groups and the reference group using the Student’s T-test. Details of all the comparisons are available in the S1 File.
Fig 3.
Differentially expressed miRNAs among individuals with different types of dysglycemia.
Volcano plots display the differentially expressed miRNA based on adjusted p-value and fold-difference variation of the miRNA expression in comparisons between each clinical group and the reference group (normoglycemic healthy controls) as follows: T1DM (A), T2DM (B) and pre-DM (C). In (D), Venn diagram shows the miRNA with significant adjusted p-values (<0.05) of each comparison between the indicated groups and the reference group using the Student’s T-test. Details of all the comparisons are available in the S1 File.
Fig 4.
Pathway Enrichment Analysis of the differentially expressed genes per each clinical group.
The differentially expressed genes (DEGs) were analyzed using a pathway enrichment compared to the KEGG database as described in the Methods section. Only statistically significant enriched pathways are depicted. No pathway was observed in the T1DM group because there were no DEGs identified.
Fig 5.
Multi-Omics Factor Analysis identified latent factors able to distinguish pre-DM and T2DM from T1DM and normoglycemic healthy controls.
(A) 46 samples with paired data on mRNA and miRNA expression were used using MOFA as described in Methods. (B) MOFA summarized the mRNA and miRNA data in 6 latent factors (LF) with different associations (evaluated using the proportion of total variance explained, R2) with the mRNA dataset, the miRNA dataset or both. (C) Each latent factor was inputted as a principal component in a PCA algorithm and a matrix was used to show different combinations of latent factors able to segregate the distinct clinical groups.
Fig 6.
Functional analysis of loading values from latent factor 1.
(A) Pathway enrichment analysis of the latent factor 1. (B) Loading values miRNA composing the latent factor 1. (C) Loading values mRNA composing the latent factor 1. In (B) and (C) the “+” and “-” signs infer directionality of the influence of the distribution of data points shown in Fig 4C.