Identification of Important Genes of childhood autism and Construction of the Diagnostic Model

Abstract

Abstract Objective To investigate the potential role of genes of inflammation in Autism spectrum disorder(ASD)and construct a model for the diagnosis of ASD. Methods In this study, transcriptome-wide profiling datasets, GSE111175, GSE18123 and GSE6575 were downloaded from Gene Expression Omnibus (GEO) database. Significant immune-related genes were identified separately to be the biomarkers for the diagnosis of ASD by using support vector machine model (SVM), RF-OOB algorithm, and LASSO regression. Results By SVM, RF-OOB and LASSO Regression screening were used to select the six key immune-related genes (ADIPOR1, CD79B,CSF2RA, HLA-DMA, HLA-DQA1, NRAS) to diagnose ASD. A nomogram model was constructed to predict ASD based on the six key immune-related genes by using “rms” package. The relative proportion of 28 immune cell types were calculated by using ssGSEA algorithm. In eight significantly different immune cells, The proportion of Macrophage, Immature Mast cell, Macrophage, Immature Mast cell, T follicular helper cell, Neutrophil, Plasmacytoid dendritic cell increased in proportion, while the ratio of Memory B cell, Activated B cell, and B cell were decreased in ASD compared to control groups were observed. Conclusions This study provides clues about the relationship between inflammation and ASD, and suggests that inflammation may be the cause of ASD and a potential therapeutic target of ASD. Through the key genes screened above, small chemical molecules directly associated with ASD disease were identified, It may be helpful to study the pathogenesis of ASD.