Machine Learning Study of SNPs in Noncoding Regions to Predict Non-small Cell Lung Cancer Susceptibility

Area of Science:

• Genetics
• Oncology
• Bioinformatics

Background:

• Non-small cell lung cancer (NSCLC) is the most prevalent lung cancer subtype.
• Both environmental and genetic factors influence lung cancer susceptibility.
• Understanding genetic risk factors is crucial for early detection and prevention.

Purpose of the Study:

• To conduct a genome-wide association study (GWAS) to identify novel genetic variants associated with NSCLC risk in a Chinese population.
• To develop a predictive model for NSCLC susceptibility using genetic and clinical data with machine learning.
• To explore the application of machine learning in precision medicine for NSCLC.

Main Methods:

• Genome-wide association study (GWAS) of 287 NSCLC patients and 467 healthy controls using Illumina Genome-Wide Asian Screening Array Chip (712,095 SNPs).
• Logistic regression modeling to identify significant single nucleotide polymorphism (SNP) loci associated with NSCLC risk.
• Machine learning algorithms applied to a combined dataset of identified SNPs, previously reported SNPs, and clinical covariates (smoking status, age, CT screening, sex).

Main Results:

• GWAS identified 17 novel noncoding region SNP loci associated with NSCLC risk, with top SNPs (rs80040741, rs9568547, rs6010259) achieving stringent p-values (<3.02e-6).
• Two top SNPs, rs80040741 and rs6010259, were intronic variants within MUC3A and MLC1 genes, respectively.
• A machine learning model integrating genetic and clinical data achieved 86% accuracy in distinguishing NSCLC patients from healthy controls.

Conclusions:

• This study identified novel genetic loci contributing to NSCLC risk in the Chinese population.
• The integration of genetic and clinical data with machine learning offers a promising approach for NSCLC early diagnosis.
• Findings enhance the understanding of machine learning applications in precision medicine for cancer susceptibility.