Limits...
Combinatorial chromatin modification patterns in the human genome revealed by subspace clustering.

Ucar D, Hu Q, Tan K - Nucleic Acids Res. (2011)

Bottom Line: We identify 843 combinatorial patterns that recur at >0.1% of the genome.A total of 19 chromatin modifications are observed in the combinatorial patterns, 10 of which occur in more than half of the patterns.We also identify combinatorial modification signatures for eight classes of functional DNA elements.

View Article: PubMed Central - PubMed

Affiliation: Department of Internal Medicine, University of Iowa, Iowa City, 52242 Iowa, USA.

ABSTRACT
Chromatin modifications, such as post-translational modification of histone proteins and incorporation of histone variants, play an important role in regulating gene expression. Joint analyses of multiple histone modification maps are starting to reveal combinatorial patterns of modifications that are associated with functional DNA elements, providing support to the 'histone code' hypothesis. However, due to the lack of analytical methods, only a small number of chromatin modification patterns have been discovered so far. Here, we introduce a scalable subspace clustering algorithm, coherent and shifted bicluster identification (CoSBI), to exhaustively identify the set of combinatorial modification patterns across a given epigenome. Performance comparisons demonstrate that CoSBI can generate biclusters with higher intra-cluster coherency and biological relevance. We apply our algorithm to a compendium of 39 genome-wide chromatin modification maps in human CD4(+) T cells. We identify 843 combinatorial patterns that recur at >0.1% of the genome. A total of 19 chromatin modifications are observed in the combinatorial patterns, 10 of which occur in more than half of the patterns. We also identify combinatorial modification signatures for eight classes of functional DNA elements. Application of CoSBI to epigenome maps of different cells and developmental stages will aid in understanding how chromatin structure helps regulate gene expression.

Show MeSH

Related in: MedlinePlus

Performance comparison of EDISA and CoSBI. (A) Intra-cluster similarity distributions of EDISA and CoSBI biclusters. (B) Hyper-geometric P-value distributions for enhancer enrichment of EDISA and CoSBI biclusters.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3105409&req=5

Figure 2: Performance comparison of EDISA and CoSBI. (A) Intra-cluster similarity distributions of EDISA and CoSBI biclusters. (B) Hyper-geometric P-value distributions for enhancer enrichment of EDISA and CoSBI biclusters.

Mentions: Using the parameters described above, EDISA produced 234 biclusters that on average include 10 genomic loci and CoSBI produced 249 biclusters that on average include 52 genomic loci. To assess the quality of the identified biclusters, we calculated their intra-cluster similarities. As can be seen from Figure 2A, the average intra-cluster similarities were 0.69 and 0.83 for EDISA and CoSBI biclusters, respectively. Lower intra-cluster similarity of EDISA biclusters suggests that the greedy algorithm had difficulty grouping genomic loci with a coherent modification signal into the same bicluster. In addition, the small size of EDISA biclusters indicates that they do not involve complete coherent patterns.Figure 2.


Combinatorial chromatin modification patterns in the human genome revealed by subspace clustering.

Ucar D, Hu Q, Tan K - Nucleic Acids Res. (2011)

Performance comparison of EDISA and CoSBI. (A) Intra-cluster similarity distributions of EDISA and CoSBI biclusters. (B) Hyper-geometric P-value distributions for enhancer enrichment of EDISA and CoSBI biclusters.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3105409&req=5

Figure 2: Performance comparison of EDISA and CoSBI. (A) Intra-cluster similarity distributions of EDISA and CoSBI biclusters. (B) Hyper-geometric P-value distributions for enhancer enrichment of EDISA and CoSBI biclusters.
Mentions: Using the parameters described above, EDISA produced 234 biclusters that on average include 10 genomic loci and CoSBI produced 249 biclusters that on average include 52 genomic loci. To assess the quality of the identified biclusters, we calculated their intra-cluster similarities. As can be seen from Figure 2A, the average intra-cluster similarities were 0.69 and 0.83 for EDISA and CoSBI biclusters, respectively. Lower intra-cluster similarity of EDISA biclusters suggests that the greedy algorithm had difficulty grouping genomic loci with a coherent modification signal into the same bicluster. In addition, the small size of EDISA biclusters indicates that they do not involve complete coherent patterns.Figure 2.

Bottom Line: We identify 843 combinatorial patterns that recur at >0.1% of the genome.A total of 19 chromatin modifications are observed in the combinatorial patterns, 10 of which occur in more than half of the patterns.We also identify combinatorial modification signatures for eight classes of functional DNA elements.

View Article: PubMed Central - PubMed

Affiliation: Department of Internal Medicine, University of Iowa, Iowa City, 52242 Iowa, USA.

ABSTRACT
Chromatin modifications, such as post-translational modification of histone proteins and incorporation of histone variants, play an important role in regulating gene expression. Joint analyses of multiple histone modification maps are starting to reveal combinatorial patterns of modifications that are associated with functional DNA elements, providing support to the 'histone code' hypothesis. However, due to the lack of analytical methods, only a small number of chromatin modification patterns have been discovered so far. Here, we introduce a scalable subspace clustering algorithm, coherent and shifted bicluster identification (CoSBI), to exhaustively identify the set of combinatorial modification patterns across a given epigenome. Performance comparisons demonstrate that CoSBI can generate biclusters with higher intra-cluster coherency and biological relevance. We apply our algorithm to a compendium of 39 genome-wide chromatin modification maps in human CD4(+) T cells. We identify 843 combinatorial patterns that recur at >0.1% of the genome. A total of 19 chromatin modifications are observed in the combinatorial patterns, 10 of which occur in more than half of the patterns. We also identify combinatorial modification signatures for eight classes of functional DNA elements. Application of CoSBI to epigenome maps of different cells and developmental stages will aid in understanding how chromatin structure helps regulate gene expression.

Show MeSH
Related in: MedlinePlus