Cancer Classification Through p53 Hotspot Mutations: An Ensemble Learning Approach.

Area of Science:

• Oncology
• Genetics
• Bioinformatics

Background:

• The tumor suppressor protein p53 plays a critical role in cancer cell survival and apoptosis.
• TP53 is the most frequently altered tumor suppressor gene in human cancers, with mutations often occurring in the DNA binding domain.
• Hotspot mutations in TP53 lead to loss of wild-type function and acquisition of oncogenic properties, promoting cancer progression and drug resistance.

Purpose of the Study:

• To classify cancer types with high accuracy and precision using TP53 mutation data.
• To investigate the role of p53 protein stability and hotspot codons in cancer classification.
• To leverage clinically and biologically meaningful features of TP53 mutations for cancer type identification.

Main Methods:

• Utilized an ensemble approach for cancer type classification based on labeled mutation data.
• Focused on six specific hotspot codons (Arg175, Gly245, Arg249, Arg248, Arg273, Arg282) of the p53 protein.
• Employed the Extreme Gradient Boosting (XGBoost) classifier for analysis.

Main Results:

• Achieved a classification accuracy of 99.85% for cancer types.
• Demonstrated high precision (99.80%), MCC (99.85%), and F1 score (99.80%).
• Attained an area under the curve (AUC) of 100%, indicating excellent model performance.

Conclusions:

• The study successfully classified cancer types with exceptional accuracy using TP53 hotspot mutation data.
• The findings highlight the potential of using p53 mutation profiles for precise cancer subtyping.
• The developed ensemble approach, particularly XGBoost, is highly effective for cancer classification based on genetic alterations.