Machine learning combined with immune cell infiltration to analyze the mechanism of ferroptosis in chronic myeloid leukemia
[Objective]The key genes of ferroptosis in chronic myeloid leukemia(CML)were screened based on machine learning method,and the potential mechanism of the key genes in CML was analyzed.[Method]Two CML datasets(GSE5550 and GSE24739)in the GEO database were integrated to eliminate the batch effect;The"LIMMA"R package was used to dif-ferentially analyze the ferroptosis gene set in 25 CML and 16 control samples;GO and KEGG enrichment analyses were per-formed for differentially differentiated genes using the"ClusterProfiler"R package;Two key genes were screened using SVM-RFE and LASSO regression;The MiRNA-TF-mRNA regulatory network of key genes was constructed by Cytoscape soft-ware;The CIBERSORT algorithm was used to analyze the composition patterns of 22 immune cell components in cml and the correlation between key genes and immune cells.[Result]A total of 34 ferroptosis differential genes were obtained.A total of 29 pathways and 388 GO entries were enriched from 34 differential genes.Machine learning identified AKR1C1,DUSP1,FADS2 and SLC1A4 as key genes for CML and validated them by Receiver Operating Characteristic Curve(ROC).The MiRNA-TF-mRNA regulatory network showed that multiple MiRNAs could simultaneously regulate ferroptosis,and SLC1A4 mainly regulated six transcription factors,CLI3,CLI2,PCBP2,BRCA2,SP2 and ZBTB17,while other transcription factors were regulated by MiR-NA.Immune cell infiltration analysis showed that SLC1A4 was closely related to T follicular helper cells.FADS2 is associated with B progenitor cells,B memory cells,CD8 T cells,T follicular helper cells,NK resting cells,macrophages M0,dendritic cells,and neutrophils.DUSP1 is closely associated with B memory cells.[Conclusion]The four key ferroptosis genes in CML,AKR1C1,DUSP1,FADS2 and SLC1A4,can be used as potential biomarkers for CML,and their association with immune cell in-filtration may provide new insights for understanding the development of CML.
NETL
NSTL
万方数据
