Skip to main content
. 2025 Oct 2;16:1642817. doi: 10.3389/fpsyt.2025.1642817

Figure 3.

A series of charts and diagrams depicting the analysis of genetic data related to CON and ADHD groups. Panel A shows plots of partial likelihood deviance and coefficients against Log Lambda. Panel B illustrates variable importance using a bar plot. Panel C presents a Venn diagram comparing LASSO and Boruta methods. Panel D contains ROC curves with AUC values for CD180 and COA3. Panel E shows box plots of expression levels for CD180 and COA3. Panel F is a scatter plot with a linear regression of CD180 versus COA3. Panel G displays a circular diagram mapping genes on chromosomes.

Identification and validation of CD180 and COA3 as ADHD biomarkers. (A) LASSO Regression for Biomarker Selection. Identifcation of candidate diagnostic biomarkers through machine learning algorithms. From an initial set of 10 hub genes, the LASSO method identified three candidate genes with a log (λ. min) of -3.355825. (B) The Boruta algorithm evaluated gene importance against randomized "shadow" features (gray). Five genes, including CD180 and COA3 (green boxes), were confirmed as significant. (C) Intersection of LASSO and Boruta Results. From LASSO and Boruta, two genes CD180 and COA3 were identified as potential biomarkers for ADHD. (D) ROC Analysis of Diagnostic Accuracy. Both biomarkers exhibited high diagnostic accuracy, achieving area under the curve (AUC) values exceeding 0.8. (E) Differential Gene Expression. Gene expression analysis revealed that CD180 and COA3 were significantly upregulated in ADHD. (F) Correlation Between Biomarkers. Correlation analysis further indicated a strong positive correlation between the expression levels of CD180 and COA3. (G) Chromosomal Localization. Chromosomal localization analysis indicated that CD180 was located on chromosome 5, while COA3 is situated on chromosome 17.