Novel risk genes for systemic lupus erythematosus predicted by random forest classification

Area of Science:

• Genetics
• Immunology
• Computational Biology

Background:

• Genome-wide association studies (GWAS) have identified some genetic risk loci for Systemic Lupus Erythematosus (SLE), but a significant portion of the genetic contribution remains unexplained.
• Identifying novel SLE risk genes and understanding their regulatory mechanisms is crucial for improving risk prediction and developing targeted therapies.

Purpose of the Study:

• To identify novel risk genes for SLE and predict individual SLE risk using a machine learning approach.
• To investigate the regulatory role of identified genetic variants and their impact on gene expression in immune cells.

Main Methods:

• A random forest classifier was developed using SNP genotype data from 1,160 SLE patients and 2,711 controls, genotyped on the Immunochip.
• Gene importance scores from the classifier were used to identify potential novel risk genes.
• Allele-specific gene expression analysis and RNA-sequencing were employed to assess the functional impact of identified variants and gene expression patterns in immune cells.

Main Results:

• The random forest classifier identified 15 potential novel SLE risk genes, including ZNF804A, CDK1, and MANF, which were not previously associated with autoimmunity.
• The classifier achieved high accuracy (AUC=0.94) in predicting patients at risk for lupus nephritis.
• Cis-regulatory SNPs affecting the expression of six top-ranked genes were identified, suggesting a regulatory role for these variants.
• The top 40 predicted genes were significantly overrepresented for differential expression in B and T cells, particularly B cells.

Conclusions:

• Random forest classification is a powerful tool for identifying novel SLE risk genes and predicting disease risk, including lupus nephritis.
• Novel candidate genes such as ZNF804A, CDK1, and MANF warrant further investigation for their role in SLE pathogenesis.
• Identified cis-regulatory variants and differential gene expression in immune cells highlight potential mechanisms underlying SLE genetic susceptibility.