Screening and Predicting Multi-Omics T-ALL Core Genes Based on PU Learning.

Area of Science:

• Oncology
• Bioinformatics
• Genomics

Background:

• T-cell acute lymphoblastic leukemia (T-ALL) is a significant neoplastic disease.
• Identifying key genes is crucial for understanding T-ALL pathogenesis and developing targeted therapies.

Purpose of the Study:

• To screen for and identify core genes associated with T-ALL.
• To develop and validate a bioinformatics approach for T-ALL gene discovery.

Main Methods:

• Integrated analysis of RNA-seq, CTCF ChIP-seq, and DNA methylation datasets.
• Application of Positive-unlabeled (PU) bagging with a multi-layer perceptron (MLP) classifier for candidate gene screening.
• Construction of a protein-protein interaction (PPI) network and functional enrichment analysis (GO, KEGG) for core gene validation.

Main Results:

• Successfully screened for differentially expressed genes (DEGs) and identified candidate T-ALL related genes.
• Utilized a PU learning approach to handle datasets with limited positive samples.
• Validated the identified core genes through functional enrichment and literature review, confirming their relevance to T-ALL.

Conclusions:

• The proposed bioinformatics method effectively predicts T-ALL related core genes.
• The identified core genes hold significant potential for future T-ALL biomarker studies and therapeutic target development.