Fig 1.
A schematic workflow of a development MetGEMs toolbox.
The computational framework for the developed MetGEMs toolbox is divided into five different sections, namely 1) the assessed and evaluated GEMs, 2) the constructed MetGEMs network as reference database towards implementing MetGEMs toolbox, 3) the validated MetGEMs toolbox’s prediction capability with shotgun metagenomic sequencing data, 4) using MetGEMs toolbox for assigning enzyme functions and relevant functional categories related human gut microbiome, and 5) annotating putative enzyme functions related in allergic disease using the MetGEMs toolbox as described in following.
Fig 2.
A metabolic diversity of assessed most up-to-date GEMs.
(A) A Pearson’s correlation analysis between KO IDs and EC numbers identified in the evaluated 818 GEMs across different taxonomic classes at a phylogenetic-tree scale. Full details of correlation analysis and goodness of fit between KO IDs and EC numbers by scatter plot can be seen in S1 Fig. Note: In this Figure, the blue and green lines indicate the numbers of KO IDs and EC numbers, respectively. PCC stands for Pearson’s correlation coefficient. The phylogenetic-tree scale is built from 773 genomes with coding sequences available in AGORA collections visualized by iTOL [26]. (B) A bar graph explicates the selected taxonomic classes in context of the average numbers of KO IDs and EC numbers. (C) A bar graph illustrates a comparison of Core- and Pan-Function computational approaches in terms of the average numbers of KO IDs across different genera.
Table 1.
The coverage list of KO IDs and EC numbers identified in MetGEMs networks underlying four different computational approaches.
Table 2.
The coverage list of predicted KO IDs and EC numbers identified by MetGEMs toolbox.
Fig 3.
Validated results of MetGEMs toolbox and its prediction capability for metabolic function inferences.
(A) A Spearman’s correlation coefficient (SCC) graph between MetGEMs toolbox’s predicted results compared with shotgun metagenomic results of corresponding samples. Note: A correlation coefficient was analyzed for both predicted KO IDs and EC numbers with 61 selected samples. (B) A horizontal bar chart of predicted KO IDs and functionally metabolic categorized by MetGEMs toolbox.
Fig 4.
MetGEMs toolbox predicted metabolic functions and routes from Thai population-based allergy birth cohort study.
Note: Boxplot represents relative abundances of enzyme functions predictions (KO IDs and EC numbers) between healthy and atopic dermatitis samples during 9–19 months. Wilcoxon rank-sum test is used for statistical significance. * and ** correspond to p-value < 0.01 and p-value< 0.005, respectively. (A) The predicted KO IDs including triosephosphate isomerase (K01803), undecaprenyl-diphosphatase (K06153), ribulose-phosphate 3-epimerase (K01783), aspartate carbamoyltransferase (K00608), aspartate carbamoyltransferase catalytic subunit (K00609). (B) The predicted EC numbers including undecaprenyl-diphosphate phosphatase (EC: 3.6.1.27), triose-phosphate isomerase (EC: 5.1.3.1), aspartate carbamoyltransferase (EC: 2.1.3.2), amidophosphoribosyltransferase (EC: 2.4.2.14), triose-phosphate isomerase (EC: 5.3.1.1) (C) The predicted metabolic routes including Superpathway of pyrimidine deoxyribonucleoside salvage (PWY-7200), L-arginine biosynthesis III (via N-acetyl-L-citrulline) (PWY-5154), L-ornithine biosynthesis I (GLUTORN-PWY), Superpathway of pyrimidine nucleobases salvage (PWY-7208), and Superpathway of sulfate assimilation and cysteine biosynthesis (SULFATE-CYS-PWY).
Fig 5.
MetGEMs toolbox identified putative enzyme functions involved in L-arginine biosynthesis III (via N-acetyl-L-citrulline) associated with atopic dermatitis.
Note: Log2 Foldchange shows the ratio of abundance difference of putative enzymes (EC numbers) between atopic dermatitis and healthy samples. Red color represents that atopic dermatitis samples have higher relative abundance of EC numbers than healthy samples. Blue color represents that healthy samples have higher relative abundance of EC numbers than atopic dermatitis samples.