Understanding Biases in Developing Molecular Signatures for Viral Infections
Author Information
Author(s): Lytkin Nikita I., McVoy Lauren, Weitkamp Jörn-Hendrik, Aliferis Constantin F., Statnikov Alexander
Primary Institution: New York University School of Medicine
Hypothesis
How do biases in data analysis affect the development of clinical-grade molecular signatures for acute respiratory viral infections?
Conclusion
The study highlights the importance of understanding and mitigating biases in data analysis to improve the development of clinically robust molecular signatures.
Supporting Evidence
- The study identified 3,473 novel non-redundant molecular signatures for acute respiratory viral infections.
- Data preprocessing methods were shown to bias gene selection, leading to an increase in false positives.
- Only 20 out of 30 genes in a previously developed signature were found to be non-redundant.
Takeaway
This study shows that when scientists analyze data to create tests for diseases, they need to be careful about mistakes that can happen during the analysis, which can lead to wrong conclusions.
Methodology
The study used a data-analytic protocol to examine biases related to signature multiplicity, biomarker redundancy, data preprocessing, and validation of signature reproducibility.
Potential Biases
Potential biases in data analysis could lead to incorrect conclusions about the effectiveness of molecular signatures.
Limitations
The study's findings may not generalize to all types of molecular signatures or diseases beyond acute respiratory viral infections.
Participant Demographics
Participants included healthy individuals and those with symptomatic acute respiratory viral infections, with a total of 112 gene expression profiles analyzed.
Statistical Information
P-Value
p<0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website