Limits...
Identification of under-detected periodicity in time-series microarray data by using empirical mode decomposition.

Chen CR, Shu WY, Chang CW, Hsu IC - PLoS ONE (2014)

Bottom Line: However, time-series microarray data are noisy.By validating a protein complex coexpression analysis, we revealed that 56 genes were newly determined as periodic.This approach can be applied to other time-series microarray studies.

View Article: PubMed Central - PubMed

Affiliation: Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Hsinchu, Taiwan.

ABSTRACT
Detecting periodicity signals from time-series microarray data is commonly used to facilitate the understanding of the critical roles and underlying mechanisms of regulatory transcriptomes. However, time-series microarray data are noisy. How the temporal data structure affects the performance of periodicity detection has remained elusive. We present a novel method based on empirical mode decomposition (EMD) to examine this effect. We applied EMD to a yeast microarray dataset and extracted a series of intrinsic mode function (IMF) oscillations from the time-series data. Our analysis indicated that many periodically expressed genes might have been under-detected in the original analysis because of interference between decomposed IMF oscillations. By validating a protein complex coexpression analysis, we revealed that 56 genes were newly determined as periodic. We demonstrated that EMD can be used incorporating with existing periodicity detection methods to improve their performance. This approach can be applied to other time-series microarray studies.

Show MeSH

Related in: MedlinePlus

Visualization of networks of nine highly coexpressed MIPS protein complexes.A colored node represents a gene and the radius of the node is proportional to the intensity or gene expression level. An edge between two nodes indicates that a gene pair is coexpressed. A red node reflects a periodically expressed gene identified by both Tu et al. and EMD. A yellow node indicates a periodically expressed gene identified by EMD, but not by Tu et al. A blue node indicates a periodically expressed gene identified by neither Tu et al. nor EMD.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4221108&req=5

pone-0111719-g007: Visualization of networks of nine highly coexpressed MIPS protein complexes.A colored node represents a gene and the radius of the node is proportional to the intensity or gene expression level. An edge between two nodes indicates that a gene pair is coexpressed. A red node reflects a periodically expressed gene identified by both Tu et al. and EMD. A yellow node indicates a periodically expressed gene identified by EMD, but not by Tu et al. A blue node indicates a periodically expressed gene identified by neither Tu et al. nor EMD.

Mentions: Coexpressed protein complexes are visualized using network graphs, in which a node indicates a gene and an undirected edge between nodes indicates a gene pair that is coexpressed (see Methods). As shown in Figure 7, the expression levels of these protein complex components were highly correlated; the more edges between nodes (genes), the higher the level of coexpression of genes encoding elements of a protein complex. The best examples of coexpression were provided by mitochondrial ribosomal complexes (Figures 7A and 7B). Tu et al. identified all the periodically expressed genes, except for one (Table 1). The expression levels of nearly all the gene pairs of cytoplasmic ribosomal complexes were highly correlated, but many of these genes were identified as non-periodic in the Tu et al. analysis (Figures 7C and 7D). The discovery of PP genes in these complexes might indicate that under-detected periodicity led to the discrepancy between the results of the original analysis and the coexpression analysis. The subcomponents of proteasome complexes were highly coexpressed. All the gene pairs encoding subcomponents of the 20S proteasome were correlated, and only two of the 153 gene pairs of the 19/22S regulator were not correlated (Table 1). We identified seven and four under-detected genes from the 19/22S regulator and the 20S proteasome complex, respectively (Figures 7E and 7F).


Identification of under-detected periodicity in time-series microarray data by using empirical mode decomposition.

Chen CR, Shu WY, Chang CW, Hsu IC - PLoS ONE (2014)

Visualization of networks of nine highly coexpressed MIPS protein complexes.A colored node represents a gene and the radius of the node is proportional to the intensity or gene expression level. An edge between two nodes indicates that a gene pair is coexpressed. A red node reflects a periodically expressed gene identified by both Tu et al. and EMD. A yellow node indicates a periodically expressed gene identified by EMD, but not by Tu et al. A blue node indicates a periodically expressed gene identified by neither Tu et al. nor EMD.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4221108&req=5

pone-0111719-g007: Visualization of networks of nine highly coexpressed MIPS protein complexes.A colored node represents a gene and the radius of the node is proportional to the intensity or gene expression level. An edge between two nodes indicates that a gene pair is coexpressed. A red node reflects a periodically expressed gene identified by both Tu et al. and EMD. A yellow node indicates a periodically expressed gene identified by EMD, but not by Tu et al. A blue node indicates a periodically expressed gene identified by neither Tu et al. nor EMD.
Mentions: Coexpressed protein complexes are visualized using network graphs, in which a node indicates a gene and an undirected edge between nodes indicates a gene pair that is coexpressed (see Methods). As shown in Figure 7, the expression levels of these protein complex components were highly correlated; the more edges between nodes (genes), the higher the level of coexpression of genes encoding elements of a protein complex. The best examples of coexpression were provided by mitochondrial ribosomal complexes (Figures 7A and 7B). Tu et al. identified all the periodically expressed genes, except for one (Table 1). The expression levels of nearly all the gene pairs of cytoplasmic ribosomal complexes were highly correlated, but many of these genes were identified as non-periodic in the Tu et al. analysis (Figures 7C and 7D). The discovery of PP genes in these complexes might indicate that under-detected periodicity led to the discrepancy between the results of the original analysis and the coexpression analysis. The subcomponents of proteasome complexes were highly coexpressed. All the gene pairs encoding subcomponents of the 20S proteasome were correlated, and only two of the 153 gene pairs of the 19/22S regulator were not correlated (Table 1). We identified seven and four under-detected genes from the 19/22S regulator and the 20S proteasome complex, respectively (Figures 7E and 7F).

Bottom Line: However, time-series microarray data are noisy.By validating a protein complex coexpression analysis, we revealed that 56 genes were newly determined as periodic.This approach can be applied to other time-series microarray studies.

View Article: PubMed Central - PubMed

Affiliation: Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Hsinchu, Taiwan.

ABSTRACT
Detecting periodicity signals from time-series microarray data is commonly used to facilitate the understanding of the critical roles and underlying mechanisms of regulatory transcriptomes. However, time-series microarray data are noisy. How the temporal data structure affects the performance of periodicity detection has remained elusive. We present a novel method based on empirical mode decomposition (EMD) to examine this effect. We applied EMD to a yeast microarray dataset and extracted a series of intrinsic mode function (IMF) oscillations from the time-series data. Our analysis indicated that many periodically expressed genes might have been under-detected in the original analysis because of interference between decomposed IMF oscillations. By validating a protein complex coexpression analysis, we revealed that 56 genes were newly determined as periodic. We demonstrated that EMD can be used incorporating with existing periodicity detection methods to improve their performance. This approach can be applied to other time-series microarray studies.

Show MeSH
Related in: MedlinePlus