Genetic Analysis of Rheumatoid Arthritis Using Random Forests
Author Information
Author(s): Sun Yan V, Cai Zhaohui, Desai Kaushal, Lawrance Rachael, Leff Richard, Jawaid Ansar, Kardia Sharon LR, Yang Huiying
Primary Institution: Department of Epidemiology, School of Public Health, University of Michigan
Hypothesis
Can random forests identify genetic associations with rheumatoid arthritis and predict disease status?
Conclusion
The study found that using the top 500 SNPs, it was possible to accurately predict RA patients and normal subjects with high sensitivity and specificity.
Supporting Evidence
- Using the top 500 SNPs, sensitivity was approximately 90% and specificity was approximately 80%.
- Candidate gene SNPs did not predict RA patients accurately when used alone.
- Random forests were used to analyze high-dimensional SNP data.
Takeaway
Scientists used a special method to look at genes to see if they can tell who has rheumatoid arthritis, and they found a way that works pretty well.
Methodology
The study used regression methods and random forests to analyze SNP data from RA patients and controls.
Potential Biases
The study's design may have led to inflation of type I error due to dependencies between data sets.
Limitations
The predictive ability of the model was not reliable across different data sets, and the genetic effects were not replicated.
Participant Demographics
839 RA cases from sib-pair families and 855 unrelated controls.
Statistical Information
P-Value
<0.0001
Confidence Interval
1.82, 3.24
Statistical Significance
p<0.05
Want to read the original?
Access the complete publication on the publisher's website