Network-Based Approach to Prioritize Results from Genome-Wide Association Studies
Author Information
Author(s): Akula Nirmala, Baranova Ancha, Seto Donald, Solka Jeffrey, Nalls Michael A., Singleton Andrew, Ferrucci Luigi, Tanaka Toshiko, Bandinelli Stefania, Cho Yoon Shin, Kim Young Jin, Lee Jong-Young, Han Bok-Ghee, McMahon Francis J.
Primary Institution: National Institute of Mental Health, National Institutes of Health, Department of Health and Human Services, Bethesda, Maryland, USA
Hypothesis
Using a network-based method can lead to a higher rate of replication in independent datasets compared to studies that rely only on single markers.
Conclusion
NIMMI effectively identifies genes involved in quantitative and categorical traits and groups them into biologically plausible networks that are highly replicable across independent studies.
Supporting Evidence
- NIMMI captured 95% of the known height associated genes within the top 20% of ranked sub-networks.
- The top 2% of NIMMI height-prioritized sub-networks were significantly enriched for genes involved in transcription and metabolism.
- NIMMI was tested on three GWAS datasets previously analyzed for height, demonstrating high reproducibility.
- NIMMI identified networks enriched for genes involved in Crohn's disease, consistent with its autoimmune nature.
Takeaway
The study created a tool called NIMMI that helps scientists find important genes by looking at how they interact with each other, making it easier to understand complex traits like height.
Methodology
NIMMI combines GWAS data with human protein-protein interaction data to create biological networks and prioritize trait-related sub-networks.
Potential Biases
Publication bias may affect the connectivity of widely studied genes compared to less studied ones.
Limitations
Some SNPs may not clearly relate to specific genes, and protein-protein interactions can be inconsistent or tissue-dependent.
Participant Demographics
The study involved participants from the InCHIANTI cohort, a population-based epidemiological study in Tuscany, Italy.
Statistical Information
P-Value
1.11×10−4
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website