Limits...
Inferring a transcriptional regulatory network of the cytokinesis-related genes by network component analysis.

Chen SF, Juang YL, Chou WK, Lai JM, Huang CY, Kao CY, Wang FS - BMC Syst Biol (2009)

Bottom Line: No literature has so far discussed the inferred results through NCA are independent of the scale of the gene expression dataset.In this study, using S. cerevisiae as a model system, NCA was successfully applied to infer similar regulatory actions of transcription factor activities from two various microarray databases and several partial transcription factor-gene connectivity datasets for selected cytokinesis-related genes independent of data sizes.Since Bud4, Iqg1, and Cdc5 are highly conserved between human and yeast, results obtained from NCA for cytokinesis in the budding yeast can lead to a suggestion that human cells should have the transcription regulator(s) as the budding yeast Mcm1-Ndd1-Fkh2 transcription factor complex in controlling occurrence of cytokinesis.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Chemical Engineering, National Chung Cheng University, Chia-Yi 62102, Taiwan. n810011@yahoo.com.tw

ABSTRACT

Background: Network Component Analysis (NCA) is a network structure-driven framework for deducing regulatory signal dynamics. In contrast to principal component analysis, which can be employed to select the high-variance genes, NCA makes use of the connectivity structure from transcriptional regulatory networks to infer dynamics of transcription factor activities. Using the budding yeast Saccharomyces cerevisiae as a model system, we aim to deduce regulatory actions of cytokinesis-related genes, using precise spatial proximity (midbody) and/or temporal synchronicity (cytokinesis) to avoid full-scale computation from genome-wide databases.

Results: NCA was applied to infer regulatory actions of transcription factor activity from microarray data and partial transcription factor-gene connectivity information for cytokinesis-related genes, which were a subset of genome-wide datasets. No literature has so far discussed the inferred results through NCA are independent of the scale of the gene expression dataset. To avoid full-scale computation from genome-wide databases, four cytokinesis-related gene cases were selected for NCA by running computational analysis over the transcription factor database to confirm the approach being scale-free. The inferred dynamics of transcription factor activity through NCA were independent of the scale of the data matrix selected from the four cytokinesis-related gene sets. Moreover, the inferred regulatory actions were nearly identical to published observations for the selected cytokinesis-related genes in the budding yeast; namely, Mcm1, Ndd1, and Fkh2, which form a transcription factor complex to control expression of the CLB2 cluster (i.e. BUD4, CHS2, IQG1, and CDC5).

Conclusion: In this study, using S. cerevisiae as a model system, NCA was successfully applied to infer similar regulatory actions of transcription factor activities from two various microarray databases and several partial transcription factor-gene connectivity datasets for selected cytokinesis-related genes independent of data sizes. The regulated action for four selected cytokinesis-related genes (BUD4, CHS2, IQG1, and CDC5) belongs to the M-phase or M/G1 phase, consistent with the empirical observations that in S. cerevisiae, the Mcm1-Ndd1-Fkh2 transcription factor complex can regulate expression of the cytokinesis-related genes BUD4, CHS2, IQG1, and CDC5. Since Bud4, Iqg1, and Cdc5 are highly conserved between human and yeast, results obtained from NCA for cytokinesis in the budding yeast can lead to a suggestion that human cells should have the transcription regulator(s) as the budding yeast Mcm1-Ndd1-Fkh2 transcription factor complex in controlling occurrence of cytokinesis.

Show MeSH
Transcriptional regulatory relationships. The transcriptional regulatory relationships between TFs (blue circles) and genes (yellow circles). Green lines indicate negative regulation. Red lines indicate positive regulation.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2800846&req=5

Figure 3: Transcriptional regulatory relationships. The transcriptional regulatory relationships between TFs (blue circles) and genes (yellow circles). Green lines indicate negative regulation. Red lines indicate positive regulation.

Mentions: The cell-cycle expression data of these 60 cytokinesis-related genes were then applied to construct the matrix E and the connective structure of the control strength matrix A through the gene-TF database, http://jura.wi.mit.edu/cgi-bin/young_public/navframe.cgi?s= 17&f, for transcriptional regulatory networks in S. cerevisiae to infer the control strength matrix and TFAs [33]. Figure 1 shows that the computational scheme for NCA, where we selected various genes out of these 60 cytokinesis-related genes, constructed the data matrix and initial structure of control strength matrix, and then deduced their values. As shown in Case I (Additional File 1: Table S2) of Figure 1, 16 genes were found to be connected to 15 TFs (see the list in Additional File 1: Table S3) in the gene-TF database. As mentioned-above, NCA requires three criteria to be satisfied in advance to ensure unique solutions for the matrix decomposition problem [16,24]. Applying the second criterion, the 15 connective TFs were used to select 592 genes from the gene-TF database. We therefore have the 592 by 18 (different time points) data matrix E and the 592 by 15 control strength matrix A. Applying the decomposition computation in the equation (5) (see methods), we yield the control strength matrix A and the 15 by 18 TFA matrix P. Figures 2 and S1 show the inferred profiles (—Š— curves) for 7 TFAs and their corresponding gene expressions. Figure 3 shows the transcriptional regulatory relationships between TFs and genes.


Inferring a transcriptional regulatory network of the cytokinesis-related genes by network component analysis.

Chen SF, Juang YL, Chou WK, Lai JM, Huang CY, Kao CY, Wang FS - BMC Syst Biol (2009)

Transcriptional regulatory relationships. The transcriptional regulatory relationships between TFs (blue circles) and genes (yellow circles). Green lines indicate negative regulation. Red lines indicate positive regulation.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2800846&req=5

Figure 3: Transcriptional regulatory relationships. The transcriptional regulatory relationships between TFs (blue circles) and genes (yellow circles). Green lines indicate negative regulation. Red lines indicate positive regulation.
Mentions: The cell-cycle expression data of these 60 cytokinesis-related genes were then applied to construct the matrix E and the connective structure of the control strength matrix A through the gene-TF database, http://jura.wi.mit.edu/cgi-bin/young_public/navframe.cgi?s= 17&f, for transcriptional regulatory networks in S. cerevisiae to infer the control strength matrix and TFAs [33]. Figure 1 shows that the computational scheme for NCA, where we selected various genes out of these 60 cytokinesis-related genes, constructed the data matrix and initial structure of control strength matrix, and then deduced their values. As shown in Case I (Additional File 1: Table S2) of Figure 1, 16 genes were found to be connected to 15 TFs (see the list in Additional File 1: Table S3) in the gene-TF database. As mentioned-above, NCA requires three criteria to be satisfied in advance to ensure unique solutions for the matrix decomposition problem [16,24]. Applying the second criterion, the 15 connective TFs were used to select 592 genes from the gene-TF database. We therefore have the 592 by 18 (different time points) data matrix E and the 592 by 15 control strength matrix A. Applying the decomposition computation in the equation (5) (see methods), we yield the control strength matrix A and the 15 by 18 TFA matrix P. Figures 2 and S1 show the inferred profiles (—Š— curves) for 7 TFAs and their corresponding gene expressions. Figure 3 shows the transcriptional regulatory relationships between TFs and genes.

Bottom Line: No literature has so far discussed the inferred results through NCA are independent of the scale of the gene expression dataset.In this study, using S. cerevisiae as a model system, NCA was successfully applied to infer similar regulatory actions of transcription factor activities from two various microarray databases and several partial transcription factor-gene connectivity datasets for selected cytokinesis-related genes independent of data sizes.Since Bud4, Iqg1, and Cdc5 are highly conserved between human and yeast, results obtained from NCA for cytokinesis in the budding yeast can lead to a suggestion that human cells should have the transcription regulator(s) as the budding yeast Mcm1-Ndd1-Fkh2 transcription factor complex in controlling occurrence of cytokinesis.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Chemical Engineering, National Chung Cheng University, Chia-Yi 62102, Taiwan. n810011@yahoo.com.tw

ABSTRACT

Background: Network Component Analysis (NCA) is a network structure-driven framework for deducing regulatory signal dynamics. In contrast to principal component analysis, which can be employed to select the high-variance genes, NCA makes use of the connectivity structure from transcriptional regulatory networks to infer dynamics of transcription factor activities. Using the budding yeast Saccharomyces cerevisiae as a model system, we aim to deduce regulatory actions of cytokinesis-related genes, using precise spatial proximity (midbody) and/or temporal synchronicity (cytokinesis) to avoid full-scale computation from genome-wide databases.

Results: NCA was applied to infer regulatory actions of transcription factor activity from microarray data and partial transcription factor-gene connectivity information for cytokinesis-related genes, which were a subset of genome-wide datasets. No literature has so far discussed the inferred results through NCA are independent of the scale of the gene expression dataset. To avoid full-scale computation from genome-wide databases, four cytokinesis-related gene cases were selected for NCA by running computational analysis over the transcription factor database to confirm the approach being scale-free. The inferred dynamics of transcription factor activity through NCA were independent of the scale of the data matrix selected from the four cytokinesis-related gene sets. Moreover, the inferred regulatory actions were nearly identical to published observations for the selected cytokinesis-related genes in the budding yeast; namely, Mcm1, Ndd1, and Fkh2, which form a transcription factor complex to control expression of the CLB2 cluster (i.e. BUD4, CHS2, IQG1, and CDC5).

Conclusion: In this study, using S. cerevisiae as a model system, NCA was successfully applied to infer similar regulatory actions of transcription factor activities from two various microarray databases and several partial transcription factor-gene connectivity datasets for selected cytokinesis-related genes independent of data sizes. The regulated action for four selected cytokinesis-related genes (BUD4, CHS2, IQG1, and CDC5) belongs to the M-phase or M/G1 phase, consistent with the empirical observations that in S. cerevisiae, the Mcm1-Ndd1-Fkh2 transcription factor complex can regulate expression of the cytokinesis-related genes BUD4, CHS2, IQG1, and CDC5. Since Bud4, Iqg1, and Cdc5 are highly conserved between human and yeast, results obtained from NCA for cytokinesis in the budding yeast can lead to a suggestion that human cells should have the transcription regulator(s) as the budding yeast Mcm1-Ndd1-Fkh2 transcription factor complex in controlling occurrence of cytokinesis.

Show MeSH