Objective To explore the mechanism of valproate-induced autism-like behavior in Wistar mice based on bioinformatics technique. Methods A total of 4 SPF grade Wistar male mice （300-350 g） and 8 female mice （200-250 g） were selected. The female and male mice were caged overnight at a ratio of 2︰1 in 8-10 weeks. The female mice were intraperitoneally injected with 600 mg/kg valproate on the 12.5 days of pregnancy. The newborn mice were included in the valproate group. Open field test， elevated cross maze test， new object recognition test and bead embedding experiment were used to evaluate the success of the model at the 35th day after birth. Another 4 pregnant mice were randomly selected to inject the same amount of normal saline once， and the newborn mice were included in the healthy control group. GSE42904 data set was used to search for genes with different expressions in the brain tissues of mice receiving valproate and methylcellulose. Genetic ontology （GO）， Kyoto Encyclopedia of Genes and Genome （KEGG）， GO/KEGG （combined with LogFC） and Gene Set Enrichment Analysis （GSEA） were used for enrichment analysis of differential genes. Quantitative real-time polymerase chain lock reaction （Q-PCR） was used to verify the expression of differential genes in the brain tissue of model mice in valproate group. Finally， 14 young mice in the valproate group and 7 young mice in the healthy control group were included. Results Compared with the control group， the time of entering the central area and the number of body extensions decreased in the open field experiment， the recognition index decreased in the new object recognition experiment， and the number of buried beads increased in the bead embedding experiment， and all the differences were statistically significant （P ＜ 0.05）. In the elevated cross maze test， there was no statistically significant difference between the valproate group and the healthy control group in the number of times of entering the open arm and the total distance of movement （P＞0.05）， indicating successful modeling. According to the analysis of GSE42904 data set， there are 267 genes in line with | Log2 （FC） | ＞ 1 and P ＜ 0.05， and these differential genes are mainly concentrated in metabolic pathway. The differential genes glutathione S-transferase α 3（Gsta 3）， gamma-glutamyl transferase 7（Ggt 7） and glucose-6-phosphate dehydrogenase （G6pd） in the glutathione metabolic pathway were verified. It was found that there was no statistical significance in the difference in mRNA expressions between Gsta 3 and Ggt 7 （P＞ 0.05）. The mRNA expression level of G6pd gene in valproate group was significantly lower than that in healthy control group， and the difference was statistically significant （P＜ 0.05）. Conclusions There are differences in the expression of G6pd gene in the valproate rat model， and the change in the expression of G6pd gene may be a pathogenic factor of autism spectrum disorder.