Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2023 Jul 17:14:1203375.
doi: 10.3389/fneur.2023.1203375. eCollection 2023.

Artificial intelligence and bioinformatics analyze markers of children's transcriptional genome to predict autism spectrum disorder

Huitao Tang #  1 , Jiawei Liang #  2 , Keping Chai  1 , Huaqian Gu  1 , Weiping Ye  1 , Panlong Cao  1 , Shufang Chen  1 , Daojiang Shen  1
Affiliations

Artificial intelligence and bioinformatics analyze markers of children's transcriptional genome to predict autism spectrum disorder

Huitao Tang et al. Front Neurol. .

Abstract

Introduction: Autism spectrum disorder (ASD), characterized by difficulties in social interaction and communication as well as restricted interests and repetitive behaviors, is extremely challenging to diagnose in toddlers. Early diagnosis and intervention are crucial however.

Methods: In this study, we developed a machine learning classification model based on mRNA expression data from the peripheral blood of 128 toddlers with ASD and 126 controls. Differentially expressed genes (DEGs) between ASD and controls were identified.

Results: We identified genes such as UBE4B, SPATA2 and RBM3 as DEGs, mainly involved in immune-related pathways. 21 genes were screened as key biomarkers using LASSO regression, yielding an accuracy of 86%. A neural network model based on these 21 genes achieved an AUC of 0.88.

Discussion: Our findings suggest that the identified neurotransmitters and 21 immune-related biomarkers may facilitate the early diagnosis of ASD. The mRNA expression profile sheds light on the biological underpinnings of ASD in toddlers and potential biomarkers for early identification. Nevertheless, larger samples are needed to validate these biomarkers.

Keywords: LASSO; RNA-Seq; autistic spectrum disorder; biomarkers; neural network.

PubMed Disclaimer

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Figures

Figure 1
Figure 1
(A) A heatmap of the clinical feature score for CON and ASD samples. (B) Distribution of hierarchical clusters 0 and 1 of ASD in PC1 and PC2 spaces. (C) Logistic regression identifies the receiver-operating characteristic (ROC) curves of hierarchical ASD-0 and ASD-1 (D) Distribution of ASD-0 and ASD-1 clusters in PC1 and PC2 spaces identified by k-means. (E) Logistic regression identifies the ROC curves of k-means ASD-0 and ASD-1. (F) The boxplot shows the difference in clinical feature scores between ASD-0 and ASD-1.
Figure 2
Figure 2
Identification of DEGs between ASD and CON and DEGs functional enrichment analysis. (A) Volcano diagram: Each point represents a protein: downregulated (blue), upregulated (red), and not significant (yellow). (B) A heatmap of the expression of DEGs in ASD and CON. (C) TOP-6 GSEA analysis of the KEGG pathway in DEGs. (D) Statistically significant Gene Ontology terms with a false discovery rate of <0.05. (E) The canonical pathway of IPA shows the signaling pathways in which DEGs are involved.
Figure 3
Figure 3
Blood feature gene selection using the LASSO binary logistic regression model. (A) LASSO coefficient profiles of the 200 DEGs. (B) A coefficient profile plot was produced against the log (lambda) sequence in the LASSO model. The optimal parameter (lambda) was selected as the first black dotted line indicated. (C) A bar graph of the coefficients of the 21 genes selected.
Figure 4
Figure 4
Construction and validation of ASD diagnosis nomogram. (A) A boxplot of the expression values of the 21 genes selected in different samples. (B) The nomogram for predicting the risk of ASD based on 21 genes. (C) The calibration curves of the ASD prediction nomogram. (D) The ROC curve of the nomogram in the validation cohort. (E) The decision curve analysis for the nomogram.
Figure 5
Figure 5
Construction and validation of the ASD diagnosis neural network. (A) Results of neural network visualization. (B) The ROC curve of the neural network in the training cohort. (C) The ROC curve of the neural network in the validation cohort. (D) The precision-recall curve for the neural network classifier.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Lai M-C, Lombardo MV, Baron-Cohen S. Autism. Lancet. (2014) 383:896–910. 10.1016/S0140-6736(13)61539-1 - DOI - PubMed
    1. Sandin S, Lichtenstein P, Kuja-Halkola R, Hultman C, Larsson H, Reichenberg A. The heritability of autism spectrum disorder. JAMA. (2017) 318:1182–4. 10.1001/jama.2017.12141 - DOI - PMC - PubMed
    1. Chang J, Gilman SR, Chiang AH, Sanders SJ, Vitkup D. Genotype to phenotype relationships in autism spectrum disorders. Nat Neurosci. (2015) 18:191–8. 10.1038/nn.3907 - DOI - PMC - PubMed
    1. Bhandari R, Paliwal JK, Kuhad A. Neuropsychopathology of autism spectrum disorder: complex interplay of genetic, epigenetic, and environmental factors. Adv Neurobiol. (2020) 24:97–141. 10.1007/978-3-030-30402-7_4 - DOI - PubMed
    1. Famitafreshi H, Karimian M. Overview of the recent advances in pathophysiology and treatment for autism. CNS Neurol Disord Drug Targets. (2018) 17:590–4. 10.2174/1871527317666180706141654 - DOI - PubMed

LinkOut - more resources