Using Random Matrix Theory to Identify Gene Co-Expression Networks
Author Information
Author(s): Luo Feng, Yang Yunfeng, Zhong Jianxin, Gao Haichun, Khan Latifur, Thompson Dorothea K, Zhou Jizhong
Primary Institution: Oak Ridge National Laboratory
Hypothesis
The universal predictions of random matrix theory can be applied to biological systems to determine correlation thresholds for gene co-expression networks.
Conclusion
The study demonstrates that random matrix theory can effectively identify gene co-expression networks and predict functions of unknown genes.
Supporting Evidence
- The application of random matrix theory revealed a sharp transition in the nearest neighbour spacing distribution of correlation matrices.
- Functional predictions for 215 unknown genes were made based on the co-expression networks.
- The co-expression network derived from yeast data was validated against known biological information.
- RMT approach was shown to be sensitive and robust against noise in the data.
- Comparison with randomized methods indicated that RMT provided more accurate identification of true links.
Takeaway
Scientists used a special math method to find out how genes work together and to guess what some unknown genes do.
Methodology
The study applied random matrix theory to analyze correlation matrices of gene expression data from various organisms.
Potential Biases
Potential biases may arise from the inherent noise in microarray data and the limited biological knowledge of certain organisms.
Limitations
The method may not work well if the number of experiments is very limited or if the data is very noisy.
Participant Demographics
Data from various organisms including yeast, E. coli, and S. oneidensis were analyzed.
Statistical Information
P-Value
0.001
Statistical Significance
p<0.001
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website