Limits...
The organization of nucleosomes around splice sites.

Chen W, Luo L, Zhang L - Nucleic Acids Res. (2010)

Bottom Line: Using the computational model of Increment of Diversity with Quadratic Discriminant (IDQD) trained from the microarray data, the nucleosome occupancy score (NOScore) was defined and applied to splice junction regions of constitutive, cassette exon, alternative 3' and 5' splicing events in the human genome.We found an interesting relation between NOScore and RNA splicing: exon regions have higher NOScores compared with their flanking intron sequences in both constitutive and alternative splicing events, indicating the stronger nucleosome occupation potential of exon regions.In addition, NOScore valleys present at approximately 25 bp upstream of the acceptor site in all splicing events.

View Article: PubMed Central - PubMed

Affiliation: Laboratory of Theoretical Biophysics, School of Physical Science and Technology, Inner Mongolia University, Hohhot 010021, China.

ABSTRACT
The occupancy of nucleosomes along chromosome is a key factor for gene regulation. However, except promoter regions, genome-wide properties and functions of nucleosome organization remain unclear in mammalian genomes. Using the computational model of Increment of Diversity with Quadratic Discriminant (IDQD) trained from the microarray data, the nucleosome occupancy score (NOScore) was defined and applied to splice junction regions of constitutive, cassette exon, alternative 3' and 5' splicing events in the human genome. We found an interesting relation between NOScore and RNA splicing: exon regions have higher NOScores compared with their flanking intron sequences in both constitutive and alternative splicing events, indicating the stronger nucleosome occupation potential of exon regions. In addition, NOScore valleys present at approximately 25 bp upstream of the acceptor site in all splicing events. By defining folding diversity-to-energy ratio to describe RNA structural flexibility, we demonstrated that primary RNA transcripts from nucleosome occupancy regions are relatively rigid and those from nucleosome depleted regions are relatively flexible. The negative correlation between nucleosome occupation/depletion of DNA sequence and structural flexibility/rigidity of its primary transcript around splice junctions may provide clues to the deeper understanding of the unexpected role for nucleosome organization in the regulation of RNA splicing.

Show MeSH

Related in: MedlinePlus

Evaluation of the performance of IDQD model. The IDQD model trained on the ChIP-chip data was compared with the SVM model and further validated on the independent high-resolution nucleosome occupancy data. (a) ROC curves for IDQD (marked with triangles) and SVM (marked with squares) models were plotted for the discrimination between nucleosome occupancy and depleted probes in the ChIP-chip data. The mean auROC of 0.958 was obtained for IDQD model in the 10-fold cross-validation experiments, higher than the SVM model with a mean auROC of 0.907. (b) An ROC curve with the auROC of 0.935 was obtained for the validation of the IDQD model in the independent nucleosome occupancy data.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2875036&req=5

Figure 1: Evaluation of the performance of IDQD model. The IDQD model trained on the ChIP-chip data was compared with the SVM model and further validated on the independent high-resolution nucleosome occupancy data. (a) ROC curves for IDQD (marked with triangles) and SVM (marked with squares) models were plotted for the discrimination between nucleosome occupancy and depleted probes in the ChIP-chip data. The mean auROC of 0.958 was obtained for IDQD model in the 10-fold cross-validation experiments, higher than the SVM model with a mean auROC of 0.907. (b) An ROC curve with the auROC of 0.935 was obtained for the validation of the IDQD model in the independent nucleosome occupancy data.

Mentions: To estimate its performance, we applied the IDQD model to the training dataset and compared its prediction quality with that of the SVM model (18) by measuring the area under the receiver operating characteristic (ROC) curve. By this metric, a random classifier achieves an area under ROC curve (auROC) of 0.5, while 1.0 corresponding to a perfect one (32). Our IDQD model obtained a mean auROC of 0.958 in the 10-fold cross-validation, superior to the SVM model with an auROC of 0.907 (Figure 1a) for the discrimination between nucleosome occupancy and depleted sequences in the same dataset.Figure 1.


The organization of nucleosomes around splice sites.

Chen W, Luo L, Zhang L - Nucleic Acids Res. (2010)

Evaluation of the performance of IDQD model. The IDQD model trained on the ChIP-chip data was compared with the SVM model and further validated on the independent high-resolution nucleosome occupancy data. (a) ROC curves for IDQD (marked with triangles) and SVM (marked with squares) models were plotted for the discrimination between nucleosome occupancy and depleted probes in the ChIP-chip data. The mean auROC of 0.958 was obtained for IDQD model in the 10-fold cross-validation experiments, higher than the SVM model with a mean auROC of 0.907. (b) An ROC curve with the auROC of 0.935 was obtained for the validation of the IDQD model in the independent nucleosome occupancy data.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2875036&req=5

Figure 1: Evaluation of the performance of IDQD model. The IDQD model trained on the ChIP-chip data was compared with the SVM model and further validated on the independent high-resolution nucleosome occupancy data. (a) ROC curves for IDQD (marked with triangles) and SVM (marked with squares) models were plotted for the discrimination between nucleosome occupancy and depleted probes in the ChIP-chip data. The mean auROC of 0.958 was obtained for IDQD model in the 10-fold cross-validation experiments, higher than the SVM model with a mean auROC of 0.907. (b) An ROC curve with the auROC of 0.935 was obtained for the validation of the IDQD model in the independent nucleosome occupancy data.
Mentions: To estimate its performance, we applied the IDQD model to the training dataset and compared its prediction quality with that of the SVM model (18) by measuring the area under the receiver operating characteristic (ROC) curve. By this metric, a random classifier achieves an area under ROC curve (auROC) of 0.5, while 1.0 corresponding to a perfect one (32). Our IDQD model obtained a mean auROC of 0.958 in the 10-fold cross-validation, superior to the SVM model with an auROC of 0.907 (Figure 1a) for the discrimination between nucleosome occupancy and depleted sequences in the same dataset.Figure 1.

Bottom Line: Using the computational model of Increment of Diversity with Quadratic Discriminant (IDQD) trained from the microarray data, the nucleosome occupancy score (NOScore) was defined and applied to splice junction regions of constitutive, cassette exon, alternative 3' and 5' splicing events in the human genome.We found an interesting relation between NOScore and RNA splicing: exon regions have higher NOScores compared with their flanking intron sequences in both constitutive and alternative splicing events, indicating the stronger nucleosome occupation potential of exon regions.In addition, NOScore valleys present at approximately 25 bp upstream of the acceptor site in all splicing events.

View Article: PubMed Central - PubMed

Affiliation: Laboratory of Theoretical Biophysics, School of Physical Science and Technology, Inner Mongolia University, Hohhot 010021, China.

ABSTRACT
The occupancy of nucleosomes along chromosome is a key factor for gene regulation. However, except promoter regions, genome-wide properties and functions of nucleosome organization remain unclear in mammalian genomes. Using the computational model of Increment of Diversity with Quadratic Discriminant (IDQD) trained from the microarray data, the nucleosome occupancy score (NOScore) was defined and applied to splice junction regions of constitutive, cassette exon, alternative 3' and 5' splicing events in the human genome. We found an interesting relation between NOScore and RNA splicing: exon regions have higher NOScores compared with their flanking intron sequences in both constitutive and alternative splicing events, indicating the stronger nucleosome occupation potential of exon regions. In addition, NOScore valleys present at approximately 25 bp upstream of the acceptor site in all splicing events. By defining folding diversity-to-energy ratio to describe RNA structural flexibility, we demonstrated that primary RNA transcripts from nucleosome occupancy regions are relatively rigid and those from nucleosome depleted regions are relatively flexible. The negative correlation between nucleosome occupation/depletion of DNA sequence and structural flexibility/rigidity of its primary transcript around splice junctions may provide clues to the deeper understanding of the unexpected role for nucleosome organization in the regulation of RNA splicing.

Show MeSH
Related in: MedlinePlus