Inferring Factors in Gene-Expression Analysis
Author Information
Author(s): Chen Bo, Chen Minhua, Paisley John, Zaas Aimee, Woods Christopher, Ginsburg Geoffrey S, Hero Alfred III, Lucas Joseph, Dunson David, Carin Lawrence
Primary Institution: Duke University
Hypothesis
Can nonparametric Bayesian methods accurately infer the number of factors in gene-expression data from viral challenge studies?
Conclusion
The proposed Bayesian models accurately inferred the number of factors in gene-expression data, identifying a significant 'pan-viral' factor associated with the host response to three viruses.
Supporting Evidence
- The Bayesian models were able to infer the number of factors accurately relative to non-Bayesian methods.
- A 'pan-viral' factor was identified that is important for early detection of viral infections.
- The study utilized gene-expression data from three different viral challenge studies.
Takeaway
This study used special math to figure out how many important factors are in gene data from people infected with viruses, helping to understand how our bodies react to these viruses.
Methodology
The study employed nonparametric Bayesian techniques, specifically the Beta Process and Indian Buffet Process, to analyze gene-expression data from three viral challenge studies.
Potential Biases
Potential biases may arise from the selection of healthy volunteers and the specific conditions of the viral challenge studies.
Limitations
The study's findings may not generalize beyond the specific viral challenges and the sample size was limited to healthy volunteers.
Participant Demographics
20 healthy volunteers participated in the rhinovirus challenge study, with additional cohorts for RSV and influenza.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website