1、Analysis of SPRITE Clusters Reveals Organizing Principles in Chromosomal InteractionYichen WANGXiaole Shirley Liu lab09/23/2018Proximity ligation based methods (3C and derivatives)Dekker et al. 2013 Nat Rev GenetTwo Strategies to Investigate Genome OrganizationLieberman-Aiden, Erez, et al. 2009 Scie

2、nceHi-CTwo Strategies to Investigate Genome OrganizationFluorescence in situ hybridizationlooping-out of the gene-dense regionsgene-rich human chromosome 19 (red) and gene-poor chromosome 18 (green)Bickmore WA, van Steensel B. 2013 Cell DNA sites close in spatial distances may still be too far apart

3、 to be directly ligated!Lieberman-Aiden, Erez, et al. 2009 ScienceSPRITE Identifies Interactions Across Larger Spatial Distances Than Can Be Observed by Hi-CSPRITE Barcode Interacting Molecules Using a Split-Pool StrategyA SPRITE cluster: all unique DNA reads that contain the same barcodeDetect inte

4、ractions across larger spatial distances than Hi-C (proximity ligation-independent)Detect multi-way interactions(barcoding)Can be expanded to detect RNA interactionSplittingPoolingTaggingRepeatn timesSequencingCross-linking,FragmentationQuinodoz et al. 2018 CellSPRITE Identifies Interactions that Oc

5、cur across Large Genomic Distances and across ChromosomeContact frequency decays with distancemore slowly in large clustersLarge SPRITE clusters harbor information for Inter-chromosomal interactionsQuinodoz et al. 2018 CellA Global View of SPRITE Detected Inter-chromosomal InteractionsNormalized con

6、tact frequencyInteractions per kb-2k-0Normalized by dividing the raw inter-chromosomal contacts by the length of chromosome and setting the sum to 1.Questions about Inter-chromosomal InteractionsWhat s a good definition of “inter-chromosomal clusters” and “Intra-chromosomal clusters”?Whats the diffe

7、rence between “inter-chromosomal clusters” and “Intra-chromosomal clusters”? eg. transcriptional factor binding, genomic featuresHow to Define Inter- and Intra- SPRITE ClustersDefinition:A SPRITE cluster: all unique DNA reads that contain the same barcodek200 clustersk500 clustersk700 clusters 0.9In

8、tra 0.9Intra 0.9IntraRegions in Intra-chromosomal Clusters Are Generally More CompactExample of an intra-clusterInteraction frequency of each 1kb regionsMouse chromosome 9 coordinate Mouse chromosome 9 coordinate Example of an inter-clusterUsing Information Entropyto Describe Different Distribution

9、of Interacting RegionsWith a given size of the cluster, H(x) reaches its maximum when regions in the cluster is evenly distributed; reaches its minimum(0) when all interactions fall into one bin.H(x) : Information entropyn: total number of es:probability of the i th eInformation entropy has a close-

10、to-linear relationship with log(k). One way of normalization to remove the effect of k is to divide it by log(k). Eliminate the Influence of k in Calculating Information EntropyR-squared: 0.9775Higher Entropy Characterize Higher Degree of Dispersion for Regions in Inter-chromosomal ClustersInterIntr

11、aInterIntrap-value 2.2e-16Inter-clusters have consistently higher entropy than intra-clusters at multiple scales, indicating that interacting regions are more dispersed in inter-clusters bin size = 1MNormalized EntropyNormalized EntropyLog10(bin size)Little Difference in TF Binding Profiles from Int

12、ra to Inter ClustersOverlapping ratio between top 10k TF peaks and each cluster (represented by a row)TF bindingratio Intra-clustersInter-clustersLittle difference in TF binding patterns for inter- and intra- chromosomal clusters.For each group, a small number of SPRITE clusters have higher binding

13、ratio for many TFs ( red lines on the bottom)Summary: Inter vs. Intra Chromosomal ClustersSPRITE clusters tend to be either involving either more than 4 chromosomes or mainly one chromosome. Tendency for the former increases when k increases.Information entropy indicates that interacting sites in in

14、tra-chromosomal clusters are more localized on chromosome than those in inter-chromosomal clusters.Next StepWhat might contribute to inter-chromosomal interactions? eg. lncRNAAverage RNA-DNA association InterIntraTF Binding Profile for Large SPRITE ClustersTF binding ratio for top 200 ChIP-seq sampl

15、esTF bindingratio POLR2- depleted clustersPOLR2- enriched clustersLayer et al. 2018 Nat MethodOverlapping ratio between top 10k TF peaks and each cluster (represented by a row)Questions about Clusters with Different TF Binding Patterns:What links the POLR2-depleted (and also TF-depleted) “cold” clus

16、ters together?Whats the genomic features of regions in these POLR2-depleted “cold” clusters ?SPRITE Clusters Have Two Different TF Binding PatternsAfter z-score normalization of TF binding ratio within each cluster (mouse embryonic stem cells):POLR2- enriched clustersPOLR2- depleted clustersZ-score

17、normalized TF binding ratio for top 200 ChIP-seq samplesZ-scoreSPRITE Clusters Have Two Different TF Binding PatternsPOLR2 enriched clustersPOLR2 depleted clustersZ-score of TF binding ratioZ-score of TF binding ratioTop TFs are different in these two kind of clusters:SPRITE Clusters Have Two Differ

18、ent TF Binding PatternsFor human, top TFs are also different in these two group of clusters:Z-score of TF binding ratioZ-score of TF binding ratioPOLR2-enriched clustersPOLR2-depleted clustersHigher Information Entropy Indicates Higher Degree of Dispersion for Regions in POLR2-enriched Clusters POLR2-enriched clusters have higher entropy POLR2-depleted clusters, indicating that interacting regions are more dispersed