Gene analysis of pituitary tumor tissue and normal tissue expression
We analyzed for genetic differences, and the flow of the study in this paper is shown in Fig. 1. Using |logFC|>0.5 as the filtering condition, DEGs analysis showed 101 DEGs, including 96 down-regulated genes and 5 up-regulated genes are shown in Fig. 1A. Upregulated genes in colorectal cancer patients include DDX28, ZNF624, SLC8A3, etc., and the down-regulated genes included POMC, TSHB, GAL, etc. are shown in Fig. 1B.
DEGs enrichment function
During GO enrichment, cell fate specification, endocrine system development, pituitary gland development, etc. were found by analysis; cellular components were aggregated in endoplasmic reticulum lumen, tight junction and collagen-containing extracellular matrix, etc.; molecular functions are clustered in fibronectin binding, bHLH transcription factor binding, insulin-like growth factor I binding, etc. (Fig. 2A); KEGG showed that DEGs were clustered in the Notch signaling pathway, Hippo signaling pathway and Human papillomavirus infection (Fig. 2B); We further performed GSEA analysis and in pituitary tumor patients the KEGG pathway clustered in chemokine signaling pathway, chronic myeloid leukemia and erbb signaling pathway (Fig. 3A-B), and entries clustered in gobp mitotic nuclear division, gobp regulation of mitotic nuclear division and gobp regulation of nuclear division, etc. (Fig. 3C-D).
LASSO and SVM- RFE regression screening of signature genes for diagnosis of pituitary tumors and validation
After LASSO regression and SVM-RFE regression yielded 11 and 17 genes for the diagnosis of pituitary tumor, respectively, the intersection was taken and 10 genes were finally screened as signature genes for the diagnosis of pituitary tumor (as shown in Fig. 4A-C). LHX3, TBX19, SOX9, CNTNAP2, RAB11FIP3, LENG8, TGFBR3, FAM150B, TMEM200B, and CRY2. The diagnostic ability of the 10 signature genes for pituitary tumors, the area under the ROC curve, was 88.3%,89.2%,93.3%,92.1%,96.7%,90.8%,97.5%,99.6%,97.9%,99.6%, respectively (as shown in Fig. 5).10 signature genes were significantly differentially expressed in the validation set GSE26966 pituitary tumors and and normal tissues with P < 0.05, where CNTNAP2, LHX3, RAB11FIP3, SOX9, TBX19 and TGFBR3 were lower expressed in patients with pituitary tumors (Fig. 6).The AUC for the diagnosis of pituitary tumors for the 10 signature genes was 89.7%,54.0%, 64.3%,72.2%,92.9%,84.1%,100.0%,93.7%,91.3%,55.6%, respectively (Fig. 7).
Immune infiltrating cells in pituitary tumors
The relative content of immune cells in pituitary tumor tissue and normal tissue (Fig. 8A), and different contents of infiltrating immune cells between the two groups were B cells memory, Plasma cells and T cells CD4 memory resting (Fig. 8B). The correlation between infiltrating immune cells (Fig. 8C),infiltrating immune cells B cells memory was positively correlated with Dendritic cells activated and Dendritic cells resting, while it was negatively correlated with T cells CD4 memory activated; T cells CD4 memory resting was negatively correlated with T cells CD8 was negatively correlated with T cells CD8 and positively correlated with Eosinophils; Plasma cells were negatively correlated with Neutrophils and positively correlated with B cells memory. Characteristic gene and immune cell correlation analysis showed in Fig. 9.