Intervene: a tool for intersection and visualization of multiple gene or genomic region sets
- 1. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Intervene: a tool for intersection and visualization of multiple gene or genomic region sets Danielle Denisko Tech Talk June 13, 2018 Template from: www.overleaf.com
- 2. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Outline Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion
- 3. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Introduction Summary: intersect and visualize sets of genes novel aspect: work specifically with genomic regions Modules: venn, upset, and pairwise
- 4. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Description Use: command line and Shiny web interface Implementation: Python 2.7 (also works with Python 3.4, 3.5, and 3.6) R Built upon: pybedtools Seaborn Matplotlib UpSetR Corrplot Venerable heatmap.2
- 5. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Installation and general usage To install: pip install intervene conda install intervene Bitbucket and Github source code Input: genomic regions in BED, GFF, or VCF format gene/name lists in plain text format
- 6. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Installation and general usage Workflow: Khan A and Mathelier A. 2017. BMC Bioinformatics. 18:287.
- 7. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Installation and general usage There are three types of output plots: Khan A and Mathelier A. 2017. BMC Bioinformatics. 18:287.
- 8. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Installation and general usage Users can provide all possible bedtools intersect options via --bedtools-options. Figure 1: There are over 15 options for specifying overlaps in bedtools intersect. Image source: BEDTools suite web page.
- 9. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Modules venn classical Venn diagram up to 6 sets input: gene lists or genomic region sets Shiny web interface provides some more flexibility: weighted and unweighted Venn and Euler diagrams different types of diagrams (up to 9 sets)
- 10. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Modules venn Figure 2: Venn (leftmost column) vs. Euler diagrams. Venn: show all 2n possible regions Euler: only show relevant (non-empty) regions Image source: Wikipedia
- 11. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Modules venn intervene venn -i RC13-KO.narrowPeak RC13-WT.narrowPeak S12-KO.narrowPeak S12-WT.narrowPeak --names=KO-rep1,WT-rep1,KO-rep2,WT-rep2 -o ~/intervene_plots/ --save-overlaps --title="RNF169 ChIP-seq peaks" --project=RNF169_KO_WT --figtype=png --figsize 12 12 --fontsize=24 --dpi=450 Figure 3: Intervene venn diagram.
- 12. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Modules venn Figure 4: Chow-Ruskey Figure 5: Edwards Figure 6: Squares Figure 7: Battle
- 13. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Modules venn Figure 8: Intervene venn diagram with 6 sets.
- 14. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Modules upset easier to interpret when there are more than 4 sets can be used effectively for 20-30 sets
- 15. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Modules upset Motivation: Figure 9: Edwards-Venn diagram for banana gene clusters comparison. D’Hont A et al. 2012. Nature. 488:7410.
- 16. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Modules upset intervene upset -i RC13-KO.narrowPeak RC13-WT.narrowPeak S12-KO.narrowPeak S12-WT.narrowPeak --names=KO-rep1,WT-rep1,KO-rep2,WT-rep2 -o ~/intervene_plots/ --figtype=png --figsize 12 12 --showshiny Figure 10: Intervene upset diagram.
- 17. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Modules upset 1 #! / usr / bin /env R s c r i p t 2 l i b r a r y ( ”UpSetR” ) 3 png ( ” I n t e r v e n e upset . png” , width =7200 , h e i g h t =3600 , 4 r e s =300) 5 e x p r e s s i o n I n p u t <− c ( ’WT−rep2 ’ =2606 , 6 ’KO−rep2 ’ =109 , 7 ’KO−rep2&WT−rep2 ’ =44, 8 ’WT−rep1 ’ =39967 , 9 ’WT−rep1&WT−rep2 ’ =12136 , 10 ’WT−rep1&KO−rep2 ’ =39, 11 ’WT−rep1&KO−rep2&WT−rep2 ’ =114 , 12 ’KO−rep1 ’ =77, 13 ’KO−rep1&WT−rep2 ’ =5, 14 ’KO−rep1&KO−rep2 ’ =21, 15 ’KO−rep1&KO−rep2&WT−rep2 ’ =22, 16 ’KO−rep1&WT−rep1 ’ =112 , 17 ’KO−rep1&WT−rep1&WT−rep2 ’ =72, 18 ’KO−rep1&WT−rep1&KO−rep2 ’ =92, 19 ’KO−rep1&WT−rep1&KO−rep2&WT−rep2 ’ =290) 20 upset ( fromExpression ( e x p r e s s i o n I n p u t ) , n s e t s =4, 21 n i n t e r s e c t s =30, show . numbers=” yes ” , 22 main . bar . c o l o r=”#ea5d4e ” , 23 s e t s . bar . c o l o r=”#317eab ” , 24 empty . i n t e r s e c t i o n s=NULL, 25 order . by = ” f r e q ” , number . a n g l e s = 0 , 26 mainbar . y . l a b e l =”No . o f I n t e r s e c t i o n s ” , 27 s e t s . x . l a b e l =” Set s i z e ” , 28 t e x t . s c a l e=c ( 2 , 2 , 2 , 2 , 2 , 3 ) ) # added to a d j u s t f o n t s i z e s 29 i n v i s i b l e ( dev . o f f ( ) )
- 18. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Modules upset Figure 11: UpSet diagram from web application.
- 19. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Modules pairwise clustered heat map of pairwise associations very large sets metrics: number of overlaps, fraction of overlap, Jaccard statistics, Fisher’s exact test, and distribution of relative distances heat map styles: tribar, dendrogram, pie, circle, square, ellipse, etc. clustering methods: various agglomerative options
- 20. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Modules pairwise Figure 12: Intervene pairwise plot.
- 21. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion ShinyApp Input: does not accept genomic regions venn: lists of names/genes/SNPs upset: lists of names/genes/SNPs, binary data, Intervene command line output listing all possible combinations of sets pairwise: lists of names/genes/SNPs, pairwise matrix of number/fraction of overlap (can be generated through Intervene on command line)
- 22. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion ShinyApp Figure 13: Screenshot from Intervene’s upset module ShinyApp.
- 23. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Plots in publications Figure 14: Coregulated and antiregulated genes (with lncRNA) in various yeast colonies. Wilkinson D et al. 2018. Oxid Med Cell Longev.
- 24. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Plots in publications Figure 15: Differentially expressed genes over time after influenza treatment. Black: up- or down-regulated genes, red: upregulated genes, blue: downregulated genes. Top: PBMCs, bottom: B cells. Jensen TL et al. 2018. F1000Research. 6:2162.
- 25. Intervene to visualize genomic region sets D. Denisko Introduction Description Installation and general usage Modules Venn diagram Upset Pairwise heatmap ShinyApp Examples Plots in publications Conclusion Conclusion Pros: simple command line tool for generating quick plots convenient for visualizing genomic region sets some customization (via output scripts and/or ShinyApp) Cons: limited ability to customize, even in ShinyApp limited plot types in comparison to ShinyApp