Flexible Visualization of the Human Phosphatome

bioRxiv preprint doi: https://doi.org/10.1101/701508; this version posted July 14, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

CoralP: Flexible visualization of the human phosphatome

Amit Min1, Erika Deoudes1, Marielle L. Bond1, Eric S. Davis2, Douglas H. Phanstiel1,2,3,4,5,6

1Thurston Arthritis Research Center, University of North Carolina, Chapel Hill, NC 27599, USA 2Curriculum in Bioinformatics & Computational Biology, University of North Carolina, Chapel Hill, NC 27599, USA 3Curriculum in Genetics & Molecular Biology, University of North Carolina, Chapel Hill, NC 27599, USA 4Department of Cell Biology & Physiology, University of North Carolina, Chapel Hill, NC 27599, USA 5Lineberger Comprehensive Cancer Research Center, University of North Carolina, Chapel Hill, NC 27599, USA 6Correspondence: [email protected]

Protein phosphatases and kinases play critical roles phosphatases, there exists a great need for methods to in a host of biological processes and diseases via the analyze, interpret, and communicate experimental results removal and addition of phosphoryl groups. While within the context of the entire protein family (Chen et al., kinases have been extensively studied for decades, 2017). recent findings regarding the specificity and activities While numerous methods have been developed to of phosphatases have generated an increased visualize the human kinome (Chartier et al., 2013; Eid et interest in targeting phosphatases for pharmaceutical al., 2017; Metz et al., 2018), no such software exists for development. This increased focus has created a the human phosphatome. Existing methods to visualize need for methods to visualize this important class of kinases make use of a phylogenetic kinase tree that was proteins within the context of the entire phosphatase constructed shortly after the sequencing of the human protein family. Here, we present CoralP, an interactive genome and later modified by Cell Signaling (www. web application for the generation of customizable, cellsignal.com). These software packages allow for the publication-quality representations of human encoding of multiple kinase attributes and data types phosphatome data. Phosphatase attributes can be through colors, sizes, and shapes laid out in a kinase tree encoded through edge colors, node colors, and node organized by sequence and functional similarity. Last sizes. CoralP is the first and currently the only tool year, we developed Coral, an interactive web application designed for phosphatome visualization and should that produces publication-quality visualizations of the be of great use to the signaling community. The source human kinome (Metz et al., 2018). Coral allows for the code and web application are available at https:// simultaneous encoding of three attributes per kinase, github.com/PhanstielLab/coralp and http://phanstiel- supports three modes of visualization, and produces high- lab.med.unc.edu/coralp respectively. resolution vector images. Recently, Chen et al. published a phylogenetic phosphatase tree (Chen et al., 2017); INTRODUCTION however, to the best of our knowledge, there are currently no available tools to visualize phosphatase data within the P Protein phosphorylation is an important post- context of this or any other format. translational modification that plays a major role in Here, we describe CoralP, a web application for regulating protein activity and function. Misregulation of the generation of customizable, publication-quality phosphorylation has been implicated in a host of human visualizations of the human phosphatome. Qualitative diseases including cancer (Ostman et al., 2006; Tonks, and quantitative features can be represented in branch 2006), rheumatoid arthritis (Begovich et al., 2004; Hendriks colors, node colors, and node sizes. Phosphatases can et al., 2013), and various neurological disorders (Pulido and be organized using the published phosphatome tree or Hooft van Huijsduijnen, 2008; Robinson and Dixon, 2005). As as radial or force-directed networks. CoralP is simple to such, altering protein phosphorylation levels has become a use, well documented, and freely available. It is the first and major focus of pharmaceutical development (Cohen, 2002). only dedicated tool for phosphatome visualization and is Protein phosphorylation is regulated primarily by protein widely applicable to a variety of data types including those kinases and protein phosphatases which add and remove generated from proteomic, genomic, epidemiological, and phosphoryl groups respectively. While protein kinases high-throughput screening experiments. have been the primary focus of research for the past several decades, recent findings regarding the activities RESULTS and specificities of their counterparts have generated an increased interest in the role of phosphatases in human CoralP is built using the same underlying framework disease research (Andersen et al., 2004; Sacco et al., as Coral and therefore makes use of similar strategies 2012; Tiganis and Bennett, 2007; Tonks, 2006). Because for data input, setting selection, and data download. of this interest and the fact that there are over 180 human Phosphatase attributes can be entered using pull-down

1 bioRxiv preprint doi: https://doi.org/10.1101/701508; this version posted July 14, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

PTPN14 PTPN13 PTPN21 Auto-generate legend PTPN20Color branches by value Match text color to branches PTPN3 PTPN1 PTPN23 PTPN2 PTPN4 PPP4C PPP6C BranchPTPN18 Color PPP2CB log2 RNA (ES / fibroblast)PTPN9 -3 0 3 PTPRS-D2 PTPRG-D2 PPP2CA PTPRD-D2 RPTPZ1-D2 PTPRN2 PTPRU-D2 PPP1CA Node Color PTPRN PTPRK-D2 PTPRF-D2 PPP1CB log2 RNA (ES / fibroblast) -3 0 3 PTPRT-D2 PPP1CC PTPRM-D2 PPP3CA PTPRA-D2 PPP3CB Node Size PTPRE-D2 PPP ES — RNA RPKM PPP3CC 0 100 FIG4 SACM1L CC1 PTPRC-D2 PPP4C PPP6C PPPL PPP2CB PTP PTPRR PTPN6 PTPN11 INPP5F PPP2CA PPP5C PAP PTPN5 PTPN14 PTPN13 PTPN21 PTPN7 SYNJ1 PPP1CA Paladin PTPN20 SAC PTPN3 PPP1CB PTPN1 SYNJ2 PTPN2 PTPN23 PPP1CC PTPN4 PPP3CA PTPN22 PTPN12PTPN18 PTPRB PTPN9 MTMR10 PPP3CB PPEF2 MTMR14 MTMR11 PTPRS-D2 PTPRG-D2 PTPRO PTPRD-D2 PPP PTPRJ RPTPZ1-D2 PPP3CC PTPRS-D1 PTPRQ PTPRN2 MTMR12 PTPRU-D2 PTPRD-D1 PPEF1 ACP5 PTPRN SBF1 PTPRK-D2 PTPRH PTPRF-D2 PTPRF-D1 ACP7 PTPRZ1-D1 PTPRT-D2 PPP5C PAP PTPRM-D2 PTPRG-D1 PTPRK-D1 PTPRA-D2 PTPRU-D1 PTPRE-D2 PPEF2 PTPRM-D1 FIG4 PGAM1 PTPRT-D1 SACM1L PTPRC-D2 PGAM2 PTPRC-D1 PPEF1 BPGM PGAM4 PALD1 PTP INPP5F HP PTPRA-D1 SYNJ1 Paladin MINPP1 TIGAR SAC SYNJ2 PTPRE-D1 SSH1 DUSP1 MTMR10 PPIP5K2 DUSP4 SSH3 DSP + MTMR14 MTMR11 PFKFB3 DUSP15 HP1 DUSP22 PTEN MTMR12 PPIP5K1 DUSP2 PFKFB2 SSH2 SBF1 MTMR6 Myotubularin SBF2 PFKFB1 DUSP5 MTMR7 PXYLP1 DUSP8 DUSP19 PFKFB4 PGAM5 DUSP16 MTMR8 HP2 DUSP12 ACP6 DUSP10 DUSP28 MTMR9 UBASH3B DUSP9 DUSP21 PPM MTMR1 UBASH3A DUSP18 ACPT DUSP7 MTMR2 ACPP MTMR4 MTM1 MTMR3 PDP1 PDP2 ACP2 PTPMT1 DUSP14 DUSP6 STYXL1 TPTE2 PHLPP1 RNGTT PHLPP2 TPTE PPTC7 STYX

Laforin DUSP11 PTEN PPM1A DUPD1 ILKAP TNS1 PPM1B DUSP13 CDKN3 HAD DUSP26 TNS3 PPM1F PPM1N DUSP3 CDC14B DUSP23 TNS2 DUSP27 Dullard CDC14A DNAJC6 PPM1E SCP3 GAK SCP2 TIMM50 PPM1G PTPDC1 SCP1 PTP4A3 PPM1D PTP4A1 PTP4A2 PP2D1 FCP1 CTDSPL2 TAB1 PPM1L PPMFCP PPM1M PPM1H PPM1J PPM1K UBLCP1 CDC25A CC3 EYA3 PDP1EYA4 PDP2 CC2 AP CECR5 PHLPP1 PGP EYA2 EYA ALPL PHLPP2 ALPI PPTC7 CIN CDC25B PHP RTR1 EYA1 NagD ALPP LMWPTP SSU72 HDHD2 CDC25C LHPP ALPPL2 PHPT1 RPAP2 PPM1A ILKAP PTPRE-D1 PTPRE-D2 PPM1B DUSP1 SSH1 SSH3 FIG4 DUSP4 DUSP15 DSP + SACM1L PTPRC-D2 PPM1F PPM1N DUSP22 PTEN DUSP2 INPP5F SSH2 PPM1E SYNJ1 DUSP5Paladin PPM1G SAC SYNJ2 DUSP8 PPM1D DUSP16 PP2D1 MTMR14 MTMR11 MTMR10 DUSP10 MTMR12 TAB1 PPM1L DUSP9 SBF1 PPM1M SBF2 MTMR6 PPM1H PPM1J PPM1K DUSP7 Myotubularin MTMR7 DUSP6 STYXL1 MTMR8 STYX MTMR9 Add color nodes esize nodes byMTMR1 value Set node opacity stroke

Figure 1. CoralP tree plot highlighting some of it’s features. The default view of a CoralP tree plot is shown in grey. Circular inserts reveal some of the formatting options. (Top left) Legends are automatically generated as users configure settings. (Top middle) Branch color can be used to encode quantitative or qualitative information. (Top right) Users can adjust the font type and color of phosphatase labels and even match the label color to the branch color. (Bottom left) Nodes can be added and recolored to depict qualitative or quantitative information. (Bottom middle) Node size can be used to encode quantitative information. (Bottom right) Users can control many aesthetic options including node opacity and stroke color.

2 bioRxiv preprint doi: https://doi.org/10.1101/701508; this version posted July 14, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

menus or pasted into text boxes. CoralP supports multiple dashboard (Chang and Borges Ribeiro, 2018), shinyBS (Bailey, identifiers including UniProt, ENSEMBL, Entrez, and HGNC. 2015) and shinyWidgets (Perrier et al., 2019) were used for the web framework. Visualization options including color palettes, data scales, fonts, and protein identifiers can all be edited via an intuitive The R packages readr (Wickham et al., 2018) and rsvg (Ooms, graphical user interface. CoralP was written using the 2018) were used for data manipulation and rendering the SVG reactive programming package Shiny (Chang et al., 2018), elements. so CoralP plots update in real-time as users adjust settings, RColorBrewer (Neuwirth, 2014) was utilized for color palettes. making it easy to generate a plot perfectly customized to the user's preferences. The Circle and Force layouts were written using the D3.js library CoralP offers highly flexible methods for visualizing (Bostock et al., 2011).

qualitative and quantitative phosphatase attributes within The phosphatome tree was adapted from Chen et al. and manually the context of the entire protein family (Figure 1). Data redrawn using vector graphics in Adobe Illustrator (Chen et al., can be encoded via multiple modalities and displayed via 2017). either tree or network views. Qualitative and quantitative phosphatase attributes can be encoded through three ACKNOWLEDGMENTS separate modalities: branch color, node color, and node We thank Bulent Arman Aksoy, Jeff Hammerbacher, Matthew E. size (for quantitative data only). Users can employ individual Berginski, and Shawn M. Gomez for their help with the infrastruc- modalities or any combination thereof. The resulting data ture required to build and host Coral and CoralP. can be displayed either as a phosphatase tree (modified from Chen et al.) or as a network organized in a radial or FUNDING force-directed fashion. For ease of use, CoralP is highly documented and freely D.H.P. is supported by the National Institutes of Health (NIH), available. An ‘Info’ page describes every feature of CoralP National Human Genome Research Institute (NHGRI) grant R00HG008662 and National Institutes of Health (NIH), National In- and includes screenshots demonstrating how different stitute of General Medical Sciences (NIGMS) grant R35GM128645. settings impact the appearance of the final plots. For maximal reproducibility and stability, the code is open- AUTHOR CONTRIBUTIONS source, version controlled, and publicly available at https:// github.com/PhanstielLab/coralp. The web application is Conceptualization: DHP available at http://phanstiel-lab.med.unc.edu/coralp. Software: AM, EMD, ESD, DHP Writing: AM, MLB, DHP DISCUSSION Visualization: EMD

To the best of our knowledge, CoralP is the first and only Supervision: DHP dedicated tool for visualizing data within the context of the Resources: DHP human phosphatome. It offers a rich selection of highly customizable features and produces high-resolution DECLARATION OF INTERESTS publication-quality figures in seconds. It is available online, The authors declare no competing interests. employs a simple graphical user interface, and includes detailed documentation and examples. As such, CoralP WEB RESOURCES is highly accessible to users, independent of operating system or computational expertise. Given its ease of use, CoralP Source Code, https://github.com/PhanstielLab/coralp the extensive adoption of the kinome tree to visualize kinome data, and the growing importance of phosphatases CoralP, http://phanstiel-lab.med.unc.edu/coralp as an area of research and drug development, we anticipate that CoralP will be of great use to the signaling community. REFERENCES CoralP will expedite phosphatase research by enhancing Begovich, Ann B., Victoria E. H. Carlton, Lee A. Honigberg, Steven our ability to interpret and communicate studies focused J. Schrodi, Anand P. Chokkalingam, Heather C. Alexander, Kristin G. on the human phosphatome. Ardlie, et al. 2004. “A Missense Single-Nucleotide Polymorphism in a Gene Encoding a Protein Tyrosine Phosphatase (PTPN22) METHODS Is Associated with Rheumatoid Arthritis.” American Journal of Human Genetics 75 (2): 330–37. CoralP is adapted from Coral and is written using R and Javascript. Chen, Mark J., Jack E. Dixon, and Gerard Manning. 2017. “Genomics The R package shiny (Chang et al., 2018) and extensions shiny- and Evolution of Protein Phosphatases.” Science Signaling 10

3 bioRxiv preprint doi: https://doi.org/10.1101/701508; this version posted July 14, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

(474). https://doi.org/10.1126/scisignal.aag1796. Kinome Render: a stand-alone and web-accessible tool to annotate the human protein kinome tree. PeerJ 1, e126. Cohen, Philip. 2002. “Protein Kinases--the Major Drug Targets of the Twenty-First Century?” Nature Reviews. Drug Discovery 1 (4): Chen, M.J., Dixon, J.E., and Manning, G. (2017). Genomics and 309–15. evolution of protein phosphatases. Sci. Signal. 10.

Hendriks, Wiljan J. A. J., Ari Elson, Sheila Harroch, Rafael Pulido, Cohen, P. (2002). Protein kinases--the major drug targets of the Andrew Stoker, and Jeroen den Hertog. 2013. “Protein Tyrosine twenty-first century? Nat. Rev. Drug Discov. 1, 309–315. Phosphatases in Health and Disease.” The FEBS Journal 280 (2): 708–30. Eid, S., Turk, S., Volkamer, A., Rippmann, F., and Fulle, S. (2017). KinMap: a web-based tool for interactive navigation through Metz, Kathleen S., Erika M. Deoudes, Matthew E. Berginski, Ivan human kinome data. BMC Bioinformatics 18, 16. Jimenez-Ruiz, Bulent Arman Aksoy, Jeff Hammerbacher, Shawn M. Gomez, and Douglas H. Phanstiel. 2018. “Coral: Clear and Hendriks, W.J.A.J., Elson, A., Harroch, S., Pulido, R., Stoker, A., and Customizable Visualization of Human Kinome Data.” Cell Systems den Hertog, J. (2013). Protein tyrosine phosphatases in health and 7 (3): 347–50.e1. disease. FEBS J. 280, 708–730.

Ostman, Arne, Carina Hellberg, and Frank D. Böhmer. 2006. Metz, K.S., Deoudes, E.M., Berginski, M.E., Jimenez-Ruiz, I., Aksoy, “Protein-Tyrosine Phosphatases and Cancer.” Nature Reviews. B.A., Hammerbacher, J., Gomez, S.M., and Phanstiel, D.H. (2018). Cancer 6 (4): 307–20. Coral: Clear and Customizable Visualization of Human Kinome Data. Cell Syst 7, 347–350.e1. Pulido, Rafael, and Rob Hooft van Huijsduijnen. 2008. “Protein Tyrosine Phosphatases: Dual-Specificity Phosphatases in Health Neuwirth, E. (2014). RColorBrewer: ColorBrewer Palettes. and Disease.” The FEBS Journal 275 (5): 848–66. Ooms, J. (2018). rsvg: Render SVG Images into PDF, PNG, Robinson, Fred L., and Jack E. Dixon. 2005. “The Phosphoinositide- PostScript, or Bitmap Arrays. 3-Phosphatase MTMR2 Associates with MTMR13, a Membrane- Associated Pseudophosphatase Also Mutated in Type 4B Charcot- Ostman, A., Hellberg, C., and Böhmer, F.D. (2006). Protein-tyrosine Marie-Tooth Disease.” The Journal of Biological Chemistry 280 phosphatases and cancer. Nat. Rev. Cancer 6, 307–320. (36): 31699–707. Perrier, V., Meyer, F., and Granjon, D. (2019). shinyWidgets: Custom Sacco, Francesca, Livia Perfetto, Luisa Castagnoli, and Gianni Inputs Widgets for Shiny. Cesareni. 2012. “The Human Phosphatase Interactome: An Pulido, R., and Hooft van Huijsduijnen, R. (2008). Protein tyrosine Intricate Family Portrait.” FEBS Letters 586 (17): 2732–39. phosphatases: dual-specificity phosphatases in health and Tonks, Nicholas K. 2006. “Protein Tyrosine Phosphatases: From disease. FEBS J. 275, 848–866. Genes, to Function, to Disease.” Nature Reviews. Molecular Cell Robinson, F.L., and Dixon, J.E. (2005). The phosphoinositide-3- Biology 7 (11): 833–46. phosphatase MTMR2 associates with MTMR13, a membrane- Andersen, J.N., Jansen, P.G., Echwald, S.M., Mortensen, O.H., associated pseudophosphatase also mutated in type 4B Charcot- Fukada, T., Del Vecchio, R., Tonks, N.K., and Møller, N.P.H. (2004). Marie-Tooth disease. J. Biol. Chem. 280, 31699–31707. A genomic perspective on protein tyrosine phosphatases: gene Sacco, F., Perfetto, L., Castagnoli, L., and Cesareni, G. (2012). The structure, pseudogenes, and genetic disease linkage. FASEB J. 18, human phosphatase interactome: An intricate family portrait. FEBS 8–30. Lett. 586, 2732–2739. Bailey, E. (2015). shinyBS: Twitter Bootstrap Components for Shiny. Tiganis, T., and Bennett, A.M. (2007). Protein tyrosine phosphatase Begovich, A.B., Carlton, V.E.H., Honigberg, L.A., Schrodi, S.J., function: the substrate perspective. Biochem. J 402, 1–15. Chokkalingam, A.P., Alexander, H.C., Ardlie, K.G., Huang, Q., Smith, Tonks, N.K. (2006). Protein tyrosine phosphatases: from genes, to A.M., Spoerke, J.M., et al. (2004). A missense single-nucleotide function, to disease. Nat. Rev. Mol. Cell Biol. 7, 833–846. polymorphism in a gene encoding a protein tyrosine phosphatase (PTPN22) is associated with rheumatoid arthritis. Am. J. Hum. Wickham, H., Hester, J., and Francois, R. (2018). readr: Read Genet. 75, 330–337. Rectangular Text Data.

Bostock, M., Ogievetsky, V., and Heer, J. (2011). D3: Data-Driven Documents. IEEE Trans. Vis. Comput. Graph. 17, 2301–2309.

Chang, W., and Borges Ribeiro, B. (2018). shinydashboard: Create Dashboards with “Shiny.”

Chang, W., Cheng, J., Allaire, J.J., Xie, Y., and McPherson, J. (2018). shiny: Web Application Framework for R.

Chartier, M., Chénard, T., Barker, J., and Najmanovich, R. (2013).