Large Scale Identification of Osteosarcoma Pathogenic Genes by Multiple Extreme Learning Machine.

Area of Science:

• Oncology
• Genetics
• Bioinformatics

Background:

• Osteosarcoma, a bone cancer primarily affecting individuals under 20, has a stagnant 5-year survival rate despite current chemotherapy and surgery treatments.
• The disease exhibits significant multigenomic heterogeneity, complicating the identification of effective therapeutic targets.
• Understanding the genetic factors driving osteosarcoma is essential for advancing treatment strategies, particularly those targeting the tumor immune microenvironment.

Purpose of the Study:

• To develop a novel, efficient computational method for identifying osteosarcoma pathogenic genes.
• To overcome the limitations of traditional gene identification methods, such as high cost and time consumption.
• To lay the groundwork for future research into osteosarcoma immune microenvironment-related treatments.

Main Methods:

• Utilized known osteosarcoma-related genes and advanced computer algorithms.
• Developed a novel method named RELM (relying on extreme learning machines).
• Fused multiple extreme learning machines (ELM) within the RELM framework.

Main Results:

• The RELM method demonstrated high reliability in identifying osteosarcoma pathogenic genes.
• Achieved an Area Under the Curve (AUC) of 0.91 in 10-cross validation.
• Obtained an Area Under the Precision-Recall Curve (AUPR) of 0.88 in 10-cross validation.

Conclusions:

• The RELM method offers a reliable and efficient approach to identifying osteosarcoma pathogenic genes.
• This computational strategy surpasses traditional methods in terms of cost and time efficiency.
• The findings provide a foundation for further exploration of gene-targeted therapies and immune microenvironment treatments for osteosarcoma.