A Methylation Diagnostic Model Based on Random Forests and Neural Networks for Asthma Identification

Abstract

Background: Asthma significantly impacts human life and health as a chronic disease. Traditional treatments for asthma have several limitations. Artificial intelligence aids in cancer treatment and may also accelerate our understanding of asthma mechanisms. We aimed to develop a new clinical diagnosis model for asthma using artificial neural networks (ANN).

Methods: Datasets (GSE85566, GSE40576, and GSE13716) were downloaded from Gene Expression Omnibus (GEO) and identified differentially expressed CpGs (DECs) enriched by Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis. Random forest (RF) and ANN algorithms further identified gene characteristics and built clinical models. In addition, two external validation datasets (GSE40576 and GSE137716) were used to validate the diagnostic ability of the model.

Results: The methylation analysis tool (ChAMP) considered DECs that were up-regulated (n =121) and down-regulated (n =20). GO results showed enrichment of actin cytoskeleton organization and cell-substrate adhesion, shigellosis, and serotonergic synapses. RF (random forest) analysis identified 10 crucial DECs (cg05075579, cg20434422, cg03907390, cg00712106, cg05696969, cg22862094, cg11733958, cg00328720, and cg13570822). ANN constructed the clinical model according to 10 DECs. In two external validation datasets (GSE40576 and GSE137716), the Area Under Curve (AUC) for GSE137716 was 1.000, and AUC for GSE40576 was 0.950, confirming the reliability of the model.

Conclusion: Our findings provide new methylation markers and clinical diagnostic models for asthma diagnosis and treatment.

Figure 1

The workflow of our study.

Figure 2

Methylation landscape of GSE85566. (a) Heat map of the top 1000 most divergent CpGs; the gradient from dark blue to yellow represented the change in expression level. (b) Results of differential expression analysis of volcano plots (asthma vs healthy). The X-axis was log(deltaBeta), and the ordinate was -log10(adj.P.Val) value; DOWN (red): DECs with down-regulated expression, UP (gray): DECs with up-regulated expression, NOT (dark blue): meaningless. (c) Heat map of DECs. Dark blue to light blue means high to low expression, green represents asthma samples, red represented healthy samples, and a clustering tree aggregated similar samples together.

Figure 3

GO and KEGG analysis results. (a) GO analysis (including molecular function, cellular component, and biological process). (b) KEGG analysis.

Figure 4

(a) The effect of the number of decision trees on the error rate. The X-axis was the number of decision trees, and the Y-axis was the error rate. The increase of trees did not affect the reduction of the error rate. (b) After the variables were entered into the random forest, the top 10 DECs were listed in order of importance according to MeanDecreaseAccuracy (left) and MeanDecreaseGini (right). (c) Hierarchical clustering results of 10 DECs in GSE85566 dataset; dark colors represent high expression, light colors represent low expression, the red band above the heat map represents normal samples, and green represents asthma samples.

Figure 5

Neural network topology. (a) Artificial neural network visualization results for the train dataset. (b) ROC results (cg05075579, cg20434422, cg03907390, cg00712106, cg05696969, cg22862094, cg11733958, cg00328720, cg13570892, and cg03325522) analysis visualization for 10-fold cross-validation method.

Figure 6

Two datasets determine neural network classification efficiency. (a) ROC result of GSE137716 dataset. (b) ROC result of GSE40576. The points marked on ROC curve are the optimal threshold points, and the values in parentheses indicate sensitivity and specificity. The AUC value was the Area Under ROC Curve, X-axis was the specificity, and Y-axis was the sensitivity. The optimal threshold was marked at the inflection point, and sensitivity and specificity were listed in parentheses.