Comparison of Pattern Detection Methods in Microarray Time Series of the Segmentation Clock
Author Information
Author(s): Dequéant Mary-Lee, Ahnert Sebastian, Edelsbrunner Herbert, Fink Thomas M. A., Glynn Earl F., Hattem Gaye, Kudlicki Andrzej, Mileyko Yuriy, Morton Jason, Mushegian Arcady R., Pachter Lior, Rowicka Maga, Shiu Anne, Sturmfels Bernd, Pourquié Olivier
Primary Institution: Stowers Institute for Medical Research
Hypothesis
Can different mathematical methods effectively identify significant patterns in gene expression profiles from microarray time series data?
Conclusion
The study demonstrates that combining several distinct mathematical approaches is valuable for identifying genes with novel transcriptional patterns.
Supporting Evidence
- Four distinct mathematical methods were applied to the same microarray time series dataset.
- All methods identified previously known cyclic genes among their top ranked candidates.
- Methods predicted novel candidate cyclic genes consistent with biological knowledge.
- Method S performed best in identifying benchmark probesets.
Takeaway
The researchers used different math methods to find patterns in gene data from mouse embryos, helping to discover new genes that work in a cycle.
Methodology
Four mathematical methods (Phase consistency, Address reduction, Cyclohedron test, and Stable persistence) were applied to analyze a microarray time series dataset.
Potential Biases
Some methods rely on assumptions about periodicity, which could introduce bias in identifying non-periodic patterns.
Limitations
The methods may not capture all types of gene expression patterns, especially those that are aperiodic.
Participant Demographics
Mouse embryos were used for the study.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website