Using Random Forests to Select SNPs for Rheumatoid Arthritis Risk Analysis
Author Information
Author(s): Daniel F Schwarz, Silke Szymczak, Andreas Ziegler, Inke R König
Primary Institution: Institut für Medizinische Biometrie und Statistik, Universität zu Lübeck
Hypothesis
Can a two-stage approach using random forests effectively identify SNPs associated with rheumatoid arthritis risk?
Conclusion
The study successfully identified candidate SNPs linked to rheumatoid arthritis using a two-stage approach with random forests.
Supporting Evidence
- The two-stage approach effectively reduced the number of SNPs for further analysis.
- Five simulated loci were identified as having significant effects on rheumatoid arthritis risk.
- Smoking, gender, and parental DR alleles were found to be important covariates.
Takeaway
Researchers used a smart method to find important genetic markers for a disease by first picking a few key ones and then checking them more closely.
Methodology
The study employed a two-stage approach where random forests were used to select SNPs, followed by logistic regression to estimate their effects.
Potential Biases
Potential overfitting due to data-dependent variable selection in the first stage.
Limitations
The study relied on simulated data and lacked a specific biological hypothesis.
Participant Demographics
1500 cases and 2000 unrelated controls, including factors like gender and smoking status.
Statistical Information
P-Value
<10-15
Statistical Significance
p<0.05
Want to read the original?
Access the complete publication on the publisher's website