Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

The UCSC Cancer Genomics Browser: update 2013

The UCSC Cancer Genomics Browser: update 2013 Published online 29 October 2012 Nucleic Acids Research, 2013, Vol. 41, Database issue D949–D954 doi:10.1093/nar/gks1008 1, 1 1 1 1 Mary Goldman *, Brian Craft , Teresa Swatloski , Kyle Ellrott , Melissa Cline , 1 1 1 1 2 Mark Diekhans , Singer Ma , Chris Wilks , Josh Stuart , David Haussler and 1, Jingchun Zhu * 1 2 Center for Biomolecular Science and Engineering and Howard Hughes Medical Institute, University of California at Santa Cruz, Santa Cruz, CA 95064, USA Received September 25, 2012; Accepted October 1, 2012 ABSTRACT the number of patients which have genomic data increased, but also the amount and type of data available The UCSC Cancer Genomics Browser (https:// per patient has grown. In addition, valuable clinical infor- genome-cancer.ucsc.edu/) is a set of web-based mation from patients and their tumors are often available tools to display, investigate and analyse cancer to researchers alongside of these genomic information. genomics data and its associated clinical informa- Despite this wealth of data, analysis of the cancer tion. The browser provides whole-genome to genome can be challenging due to the limitations in current technologies to visualise, integrate, compare and base-pair level views of several different types of analyse cancer genomics data. These data, and the conclu- genomics data, including some next-generation sions they support, must be presented in a coherent system sequencing platforms. The ability to view multiple for display and analysis as well as be accessible to the datasets together allows users to make compari- scientific and medical communities. The UCSC Cancer sons across different data and cancer types. Genomics Browser (https://genome-cancer.ucsc.edu) was Biological pathways, collections of genes, genomic developed to display these expanding data sources in an or clinical information can be used to sort, aggre- integrative, interactive and versatile way as well as help gate and zoom into a group of samples. We cur- facilitate comprehensive analysis of cancer genomics and rently display an expanding set of data from its associated clinical data (4). various sources, including 201 datasets from 22 The browser is a web-based tool to integrate, visualise TCGA (The Cancer Genome Atlas) cancers as well and analyse genomic and clinical information. as data from Cancer Cell Line Encyclopedia and Experimental measurements for multiple samples are dis- played alongside their associated clinical information. Stand Up To Cancer. New features include a com- Multiple datasets can be viewed simultaneously allowing pletely redesigned user interface with an interactive comparison across studies and between different data tutorial and updated documentation. We have also types, such as gene expression and copy number variation. added data downloads, additional clinical heatmap The browser provides interactive and dynamic views of features, and an updated Tumor Image Browser the data from whole-genome to base-pair scale resolution, based on Google Maps. New security features as well as zooming to a subset of samples. Users can inter- allow authenticated users access to private actively group samples by common clinical features such datasets hosted by several different consortia as response to chemotherapy, or by genomic signatures through the public website. that predict response to a drug. Viewing genomic data by genes allows users to easily see functional changes to the genome as well as examine trends across INTRODUCTION pathways of genes. Several statistical tools are available making it possible to obtain quantitative results dynamic- Cancer has many different molecular mechanisms to ally. Additionally, the Tumor Image Viewer, based on disrupt cellular pathways, which result in uncontrolled Google Maps, allows users to interactively view slides of cell proliferation (1–3). Fortunately, development of tumor tissue samples. high-throughput genomic technologies in recent years The browser currently contains 355 datasets corres- has greatly increased the amount of data available to re- ponding to genome-wide experiments on 71 870 samples, searchers to investigate these mechanisms. Not only have *To whom correspondence should be addressed. Tel: +1 831 459 5692; Fax: +1 831 459 1809; Email: [email protected] Correspondence may also be addressed to Jingchun Zhu. Tel: +1 831 459 5232; Fax:+1 831 459 1809; Email: [email protected] The Author(s) 2012. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/3.0/), which permits non-commercial reuse, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact [email protected]. D950 Nucleic Acids Research, 2013, Vol. 41, Database issue Table 1. Dataset summary Cancer type DNA PARADIGM Gene Protein Copy number methylation pathlette expression variation TCGA acute myeloid leukemia 2 (384) 1 (179) TCGA bladder urothelial carcinoma 1 (91) 2 (106) 5 (221) TCGA brain lower grade glioma 1 (26) 1 (27) 2 (54) 6 (608) TCGA breast invasive carcinoma 2 (994) 1 (502) 4 (2870) 1 (410) 6 (4808) TCGA cervical and endocervical SCC 5 (244) TCGA colon and rectum adenocarcinoma 1 (236) 1 (208) 1 (224) 1 (463) 4 (2256) TCGA colon adenocarcinoma 2 (498) 2 (366) 2 (894) TCGA glioblastoma multiforme 2 (370) 1 (484) 4 (1693) 1 (215) 6 (3338) TCGA head and neck squamous cell carcinoma 1 (342) 2 (555) 5 (1075) TCGA kidney renal clear cell carcinoma 2 (861) 1 (69) 4 (1222) 1 (454) 6 (3028) TCGA kidney renal papillary cell carcinoma 2 (146) 1 (16) 4 (126) 6 (340) TCGA liver hepatocellular carcinoma 2 (52) 5 (275) TCGA lung adenocarcinoma 2 (406) 1 (32) 4 (488) 6 (1374) TCGA lung squamous cell carcinoma 2 (354) 1 (136) 5 (925) 6 (1442) TCGA ovarian serous cystadenocarcinoma 1 (590) 1 (546) 4 (1761) 1 (412) 6 (3366) TCGA pancreatic adenocarcinoma 1 (36) 5 (70) TCGA prostate adenocarcinoma 1 (202) 1 (60) 5 (446) TCGA rectum adenocarcinoma 2 (178) 2 (144) 2 (334) TCGA skin cutaneous melanoma 1 (242) 1 (154) 2 (442) TCGA stomach adenocarcinoma 2 (212) 1 (58) 5 (696) TCGA thyroid carcinoma 1 (260) 1 (86) 5 (715) TCGA uterine corpus endometrioid carcinoma 2 (488) 1 (53) 3 (448) 1 (200) 6 (2312) SU2C Breast Public 1 (54) 2 (92) CCLE 2 (1934) 1 (972) Other datasets from the literature 19 (3556) 17 (2206) Number of datasets by cancer type and data type; values in parenthesis are number of samples. most of which are from The Cancer Genome Atlas project NEW DATA (TCGA, https://tcga-data.nci.nih.gov/tcga/) (5). Data on The Cancer Genome Atlas the website are updated periodically to include the latest We provide an open-access portal to view, analyse and releases from TCGA and other projects. Currently, the download public data from TCGA. We have developed browser holds 201 TCGA public-tier datasets from 22 an automated pipeline that loads and processes data TCGA cancer projects, data from the Cancer Cell Line from TCGA, allowing it to be quickly downloaded from Encyclopedia project (CCLE, http://www.broadinstitute. their servers and then displayed in the browser. This org/ccle/home) (6), and 43 other published studies. The pipeline has greatly increased the number of datasets Google-maps-based microscope slide viewer has 2433 from TCGA we can load into the browser, which has slides from TCGA. allowed us to expand the cancers we have available to A controlled access mechanism is also available for 22 cancer projects including breast, pancreas and lung private data, and currently supports the Stand Up To cancer (Table 1). Cancer breast cancer dream team (SU2C, http://www. New dataset types include GISTIC2 (9) estimated seg- standup2cancer.org/dream_teams/view/an_integrated_ mented copy number variation from the TCGA approach_to_targeting_breast_cancer_molecular_ FIREHOSE pipeline. We display next-generation RNA subtypes_and_th), I-SPY 2 TRIAL (Investigation of Illumina HiSeq gene expression data, protein expression Serial Studies to Predict Your Therapeutic Response data assayed by reverse phase protein array technology, with Imaging And moLecular Analysis 2) (7), LINCS and DNA methylation profiles measured using the (Library of Integrated Network-based Cellular Signatures) (http://www.broadinstitute.org/LINCS/) (8) Illumina Infinium HumanMethylation450 platform. We and other projects. The controlled access mechanism also have datasets showing integrated gene activity level allows authorized users to view this private data alongside inferred using the PARADIGMLITE method. public data. PARADIGM (10) is pathway analysis method to infer In the past 2 years, we have made significant changes patient- or sample-specific genetic activities by incorporat- to the browser including improving the user interface, ing curated pathway interactions with genomic measure- as well as implementing user accounts, an online tutorial ments, such as gene expression and copy number data. and better documentation. The data versioning system PARADIGMLITE, a less computationally intensive we developed this year, allows us to offer users the version of PARADIGM, integrates different types of ability to download datasets. We also released a new data on individual genes without incorporating the Tumor Image Browser, based on Google Maps, that curated pathway interactions. The gene expression and offers intuitive panning and zooming across microscope copy number input data are shown in the accompanying slide images. evidence datasets. Nucleic Acids Research, 2013, Vol. 41, Database issue D951 Figure 1. TCGA GBM datasets showing differential copy number variation, gene expression and methylation for the glioma-CpG island methylator phenotype (G-CIMP). The black box emphasizes the samples characterized as G-CIMP tumors. Copy number and DNA methylation datasets, by default, use red and blue to represent amplification and deletion, respectively. Gene expression datasets, by default, use red and green to represent over- and under-expression, respectively. For the G-CIMP clinical feature, yellow represents tumors characterized as G-CIMP and black represents tumors who are not. For the gene expression subtype clinical feature the four subtypes, from black to bright yellow, are proneural, neural, classical, and mesenchymal. (A) TCGA GBM whole genome copy number profile. (B) TCGA GBM gene expression for a select set of genes. (C) TCGA GBM DNA methylation for a select set of genes. The clinical data, which is displayed alongside the screenshot of some TCGA Glioblastoma multiforme genomic information, have also been updated with more (GBM) datasets showing differential copy number vari- readable clinical feature names and values. This allows for ation, gene expression and methylation for the easier access for users who may not be familiar with the glioma-CpG island methylator phenotype (G-CIMP). abbreviations used by TCGA. Figure 1 illustrates some of Viewing the whole genome copy number profile, users these more readable clinical feature names as well as a few can easily see that the G-CIMP tumors display distinct of the new data types from TCGA. Figure 1 is a browser copy number alterations, lacking the deletion of D952 Nucleic Acids Research, 2013, Vol. 41, Database issue chromosome 10 and amplification of chromosome 7 making the controls more obvious and intuitive, enhancing shared by the majority of the non-G-CIMP samples. the readability of drawn text, and improving the responsive- The browser also shows that G-CIMP samples are ness of the site by using more modern web technologies. highly enriched in the proneural subtype, as described in Dataset selection has been moved closer to the top of the recent publication that identifies G-CIMP samples as a the page, making it easier for users to find datasets of distinct subset of human gliomas on molecular and clinical interest. All per-dataset controls have been moved to each grounds (11). When viewing genes linked to survival individual dataset display, allowing direct configuration. outcome in GBM (12) we see that individuals with Additionally, it is now easier and more natural to find or G-CIMP tumors have different gene expression and make user genesets and signatures. The user interface for the DNA methylation profiles compared with non-G-CIMP clinical heatmap has been dramatically improved allow- tumors, and a higher survival rate after treatment, as ing intuitive configuration of visible clinical features and described in the recent publication (11). subgroup construction based on clinical and genomic data. In tandem with the updated interface, the browser now Cancer Cell Line Encyclopedia has an interactive tutorial which highlights much of the new functionality. Our software detects when a user last Our newest public data are from the CCLE, a reference used the browser and automatically opens if new features library containing primarily genetic and pharmacologic have been added to the browser since they last visited. information from a large panel of human cancer cell line Additionally, the user guide was completely updated and models (6). The CCLE datasets includes genome-wide FAQs were added. copy number and gene expression profile of the cell lines, mutations assessed in 33 genes, and pharmacological response to 24 anti-cancer drugs. These datasets are avail- New clinical heatmap features able to the public. In addition to updating the interface for the clinical heatmap, some new features were added. It is now Stand Up To Cancer public datasets from the Gray lab possible to zoom in on a subset of samples by clicking We are also displaying published copy number profile and and dragging within the clinical heatmap. This makes it gene expression levels of a collection of breast cancer cell easier to more closely examine a sample and its values. lines from Joe Gray’s lab at Oregon Health and Science Clinical features are also now automatically sorted from University in association with SU2C (13,14). Response to left to right. To modify the sort order of the features in the 77 therapeutic compounds (GI50 scores), ER and HER2 heatmap, users click to move clinical features. Since receptor status, and other phenotypic information is sample order in the genomic and associated clinical shown in the clinical data. heatmap is determined by the sort order of the clinical features, rearranging clinical features automatically triggers a vertical reordering of both heatmaps. This NEW FEATURES allows users to more easily explore phenotype and Website redesign genotype associations using the browser. Additionally, The largest new feature is a completely redesigned user subgroups are now displayed as a clinical feature, interface (Figure 2). In particular, we have focused on making it easier to sort samples by any subgrouping. Figure 2. Screenshot from user documentation highlighting features in the new user interface. Nucleic Acids Research, 2013, Vol. 41, Database issue D953 Updated genes view from ovarian serous cystadenocarcinoma, 1273 from glio- blastoma multiforme and 306 from uterine corpus Users can view genomic data on a per-gene basis, where endometrioid carcinoma. probes that map to the same gene are displayed together. To display DNA methylation data from platforms with many probes per gene, especially the Infinium FUTURE DIRECTIONS HumanMethylation450 chip, we now render probes Our plans for the next year include providing users a way ordered left to right in the direction of transcription. to bookmark a view both for themselves and to share with Used in conjunction with chromosome view, this allows others. Internal progress has already been made allowing the user to easily see which section of a gene is methylated users to visualise their own patient-identifiable or unpub- under genes view, allowing evaluation of DNA methyla- lished data in a secure manner. We will also develop new tion in different genomic contexts, such as multiple viewing capabilities for multianalyte data, allowing user to promoter sites and in gene bodies (Supplementary view copy number, gene expression, and clinical data from Figure S1). The relationship between DNA methylation the same set of samples within the same map. Finally, we and transcription is more nuanced than first predicted, are planning on displaying mutation data in a new making it important to visualize exactly where in a gene heatmap visualization. the methylation is occurring (15–17). We will continue to expand the data we display from TCGA. Additionally, we will display the new public data User accounts and website architecture from LINCS, which aims to use genome-wide expression Previously, there were many browser installations at dif- profiling to catalog the cellular consequences of small- ferent websites for groups desiring private access to pro- molecule and genetic perturbations in a breadth of tected data. To reduce engineering overhead, a controlled human cell lines. We plan to publicize and provide an access mechanism was put in place to restrict the display API for our data format. of certain data to authorized users. This mechanism relies on new backend architecture that allows us to provide the same level of security as private installations but is SUPPLEMENTARY DATA transparent to the user. As part of this, we implemented Supplementary Data are available at NAR Online: user accounts, allowing us to authenticate users. This new Supplementary Figure 1. security system currently supports the SU2C breast cancer dream team, ISPY 2 TRIAL, LINCS project, and other projects. ACKNOWLEDGEMENTS Implementing user accounts resulted in two additional The authors thank Erich Weiler, Jorge Garcia and the features. First, users who had access to multiple private UCSC Genome Browser for their support. installations can now view data from all of their projects together. This allows users to make connections across data sources. User accounts also allowed us to start FUNDING saving user genesets and signatures. As we offer more National Cancer Institute (NCI) [5U24CA143858]; new features, we hope to allow users to save other data American Association for Cancer Research (AACR) as well, including bookmarks. Users do not need to be Stand Up To Cancer (SU2C) program [SU2C-AACR- authorized to sign up for a free account or access public DT0409]; UCSF Helen Diller Family Comprehensive data; this service is open to everyone. Cancer Center. Funding for open access charge: Howard Hughes Medical Institute. Data version control and downloads When we constructed the TCGA pipeline, we concur- Conflict of interest statement. None declared. rently implemented data freezes and data version control. This was done so users know what version of REFERENCES the data they are viewing and whether the data changed since it was viewed last. Users can also download the pro- 1. Hahn,W.C. and Weinberg,R.A. (2002) Modelling the molecular cessed datasets for further analysis. These processed circuitry of cancer. Nat. Rev. Cancer, 2, 331–341. 2. Hanahan,D. and Weinberg,R.A. (2011) Hallmarks of cancer: the datasets can have advantages over the original data as next generation. Cell, 144, 646–674. they have been normalized and include clinical data 3. Vogelstein,B. and Kinzler,K.W. (2004) Cancer genes and the curation. pathways they control. Nat. Med., 10, 789–799. 4. Zhu,J., Sanborn,J.Z., Benz,S., Szeto,C., Hsu,F., Kuhn,R.M., Karolchik,D., Archie,J., Lenburg,M.E., Esserman,L.J. et al. Tumor Image Browser (2009) The UCSC Cancer Genomics Browser. Nat. Methods, 6, Our previous Tumor Image Browser, hgMicroscope, has 239–240. 5. Collins,F.S. and Barker,A.D. (2007) Mapping the cancer genome. been replaced by a completely new Google-Maps based Pinpointing the genes involved in cancer will help chart a new browser. Applying the Google Maps interface to tumor course across the complex landscape of human malignancies. Sci. images has made exploring a microscope slide easier and Am., 296, 50–57. more intuitive. Currently the browser contains 2748 6. Barretina,J., Caponigro,G., Stransky,N., Venkatesan,K., TCGA images, both tumor and normal, including 1169 Margolin,A.A., Kim,S., Wilson,C.J., Leha´ r,J., Kryukov,G.V., D954 Nucleic Acids Research, 2013, Vol. 41, Database issue Sonkin,D. et al. (2012) The cancer cell line encyclopedia enables phenotype that defines a distinct subgroup of glioma. Cancer predictive modelling of anticancer drug sensitivity. Nature, 483, Cell, 17, 510–522. 603–607. 12. Colman,H., Zhang,L., Sulman,E.P., McDonald,J.M., 7. Barker,A.D., Sigman,C.C., Kelloff,G.J., Hylton,N.M., Berry,D.A. Shooshtari,N.L., Rivera,A., Popoff,S., Nutt,C.L., Louis,D.N., and Esserman,L.J. (2009) I-SPY 2: an adaptive breast cancer trial Cairncross,J.G. et al. (2010) A multigene predictor of outcome in design in the setting of neoadjuvant chemotherapy. Clin. glioblastoma. Neuro Oncol., 12, 49–57. Pharmacol. Ther., 86, 97–100. 13. Heiser,L.M., Sadanandam,A., Kuo,W.L., Benz,S.C., 8. Garnett,M.J., Edelman,E.J., Heidorn,S.J., Greenman,C.D., Goldstein,T.C., Ng,S., Gibb,W.J., Wang,N.J., Ziyad,S., Tong,F. Dastur,A., Lau,K.W., Greninger,P., Thompson,I.R., Luo,X., et al. (2012) Subtype and pathway specific responses to anticancer Soares,J. et al. (2012) Systematic identification of genomic compounds in breast cancer. Proc. Natl Acad. Sci. USA, 109, markers of drug sensitivity in cancer cells. Nature, 483, 570–575. 2724–2729. 9. Mermel,C.H., Schumacher,S.E., Hill,B., Meyerson,M.L., 14. Neve,R.M., Chin,K., Fridlyand,J., Yeh,J., Baehner,F.L., Fevr,T., Beroukhim,R. and Getz,G. (2011) GISTIC2.0 facilitates sensitive Clark,L., Bayani,N., Coppe,J.P., Tong,F. et al. (2006) and confident localization of the targets of focal somatic A collection of breast cancer cell lines for the study of copy-number alteration in human cancers. Genome Biol., 12, R41. functionally distinct cancer subtypes. Cancer Cell, 10, 515–527. 10. Vaske,C.J., Benz,S.C., Sanborn,J.Z., Earl,D., Szeto,C., Zhu,J., 15. Curradi,M., Izzo,A., Badaracco,G. and Landsberger,N. (2002) Haussler,D. and Stuart,J.M. (2010) Inference of patient-specific Molecular mechanisms of gene silencing mediated by DNA pathway activities from multi-dimensional cancer genomics data methylation. Mol. Cell. Biol., 22, 3157–3173. using PARADIGM. Bioinformatics, 26, i237–i245. 16. Eden,S. and Cedar,H. (1994) Role of DNA methylation in the 11. Noushmehr,H., Weisenberger,D.J., Diefes,K., Phillips,H.S., regulation of transcription. Curr. Opin. Genet. Dev., 4, 255–259. Pujara,K., Berman,B.P., Pan,F., Pelloski,C.E., Sulman,E.P., 17. Jones,P.A. (2012) Functions of DNA methylation: islands, start Bhat,K.P. et al. (2010) Identification of a CpG island methylator sites, gene bodies and beyond. Nat. Rev. Genet., 13, 484–492. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nucleic Acids Research Oxford University Press

The UCSC Cancer Genomics Browser: update 2013

Download PDF
6 pages

Loading next page...
 
/lp/oxford-university-press/the-ucsc-cancer-genomics-browser-update-2013-dffdBPZ0eB

References (19)

Publisher
Oxford University Press
Copyright
The Author(s) 2012. Published by Oxford University Press.
ISSN
0305-1048
eISSN
1362-4962
DOI
10.1093/nar/gks1008
pmid
23109555
Publisher site
See Article on Publisher Site

Abstract

Published online 29 October 2012 Nucleic Acids Research, 2013, Vol. 41, Database issue D949–D954 doi:10.1093/nar/gks1008 1, 1 1 1 1 Mary Goldman *, Brian Craft , Teresa Swatloski , Kyle Ellrott , Melissa Cline , 1 1 1 1 2 Mark Diekhans , Singer Ma , Chris Wilks , Josh Stuart , David Haussler and 1, Jingchun Zhu * 1 2 Center for Biomolecular Science and Engineering and Howard Hughes Medical Institute, University of California at Santa Cruz, Santa Cruz, CA 95064, USA Received September 25, 2012; Accepted October 1, 2012 ABSTRACT the number of patients which have genomic data increased, but also the amount and type of data available The UCSC Cancer Genomics Browser (https:// per patient has grown. In addition, valuable clinical infor- genome-cancer.ucsc.edu/) is a set of web-based mation from patients and their tumors are often available tools to display, investigate and analyse cancer to researchers alongside of these genomic information. genomics data and its associated clinical informa- Despite this wealth of data, analysis of the cancer tion. The browser provides whole-genome to genome can be challenging due to the limitations in current technologies to visualise, integrate, compare and base-pair level views of several different types of analyse cancer genomics data. These data, and the conclu- genomics data, including some next-generation sions they support, must be presented in a coherent system sequencing platforms. The ability to view multiple for display and analysis as well as be accessible to the datasets together allows users to make compari- scientific and medical communities. The UCSC Cancer sons across different data and cancer types. Genomics Browser (https://genome-cancer.ucsc.edu) was Biological pathways, collections of genes, genomic developed to display these expanding data sources in an or clinical information can be used to sort, aggre- integrative, interactive and versatile way as well as help gate and zoom into a group of samples. We cur- facilitate comprehensive analysis of cancer genomics and rently display an expanding set of data from its associated clinical data (4). various sources, including 201 datasets from 22 The browser is a web-based tool to integrate, visualise TCGA (The Cancer Genome Atlas) cancers as well and analyse genomic and clinical information. as data from Cancer Cell Line Encyclopedia and Experimental measurements for multiple samples are dis- played alongside their associated clinical information. Stand Up To Cancer. New features include a com- Multiple datasets can be viewed simultaneously allowing pletely redesigned user interface with an interactive comparison across studies and between different data tutorial and updated documentation. We have also types, such as gene expression and copy number variation. added data downloads, additional clinical heatmap The browser provides interactive and dynamic views of features, and an updated Tumor Image Browser the data from whole-genome to base-pair scale resolution, based on Google Maps. New security features as well as zooming to a subset of samples. Users can inter- allow authenticated users access to private actively group samples by common clinical features such datasets hosted by several different consortia as response to chemotherapy, or by genomic signatures through the public website. that predict response to a drug. Viewing genomic data by genes allows users to easily see functional changes to the genome as well as examine trends across INTRODUCTION pathways of genes. Several statistical tools are available making it possible to obtain quantitative results dynamic- Cancer has many different molecular mechanisms to ally. Additionally, the Tumor Image Viewer, based on disrupt cellular pathways, which result in uncontrolled Google Maps, allows users to interactively view slides of cell proliferation (1–3). Fortunately, development of tumor tissue samples. high-throughput genomic technologies in recent years The browser currently contains 355 datasets corres- has greatly increased the amount of data available to re- ponding to genome-wide experiments on 71 870 samples, searchers to investigate these mechanisms. Not only have *To whom correspondence should be addressed. Tel: +1 831 459 5692; Fax: +1 831 459 1809; Email: [email protected] Correspondence may also be addressed to Jingchun Zhu. Tel: +1 831 459 5232; Fax:+1 831 459 1809; Email: [email protected] The Author(s) 2012. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/3.0/), which permits non-commercial reuse, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact [email protected]. D950 Nucleic Acids Research, 2013, Vol. 41, Database issue Table 1. Dataset summary Cancer type DNA PARADIGM Gene Protein Copy number methylation pathlette expression variation TCGA acute myeloid leukemia 2 (384) 1 (179) TCGA bladder urothelial carcinoma 1 (91) 2 (106) 5 (221) TCGA brain lower grade glioma 1 (26) 1 (27) 2 (54) 6 (608) TCGA breast invasive carcinoma 2 (994) 1 (502) 4 (2870) 1 (410) 6 (4808) TCGA cervical and endocervical SCC 5 (244) TCGA colon and rectum adenocarcinoma 1 (236) 1 (208) 1 (224) 1 (463) 4 (2256) TCGA colon adenocarcinoma 2 (498) 2 (366) 2 (894) TCGA glioblastoma multiforme 2 (370) 1 (484) 4 (1693) 1 (215) 6 (3338) TCGA head and neck squamous cell carcinoma 1 (342) 2 (555) 5 (1075) TCGA kidney renal clear cell carcinoma 2 (861) 1 (69) 4 (1222) 1 (454) 6 (3028) TCGA kidney renal papillary cell carcinoma 2 (146) 1 (16) 4 (126) 6 (340) TCGA liver hepatocellular carcinoma 2 (52) 5 (275) TCGA lung adenocarcinoma 2 (406) 1 (32) 4 (488) 6 (1374) TCGA lung squamous cell carcinoma 2 (354) 1 (136) 5 (925) 6 (1442) TCGA ovarian serous cystadenocarcinoma 1 (590) 1 (546) 4 (1761) 1 (412) 6 (3366) TCGA pancreatic adenocarcinoma 1 (36) 5 (70) TCGA prostate adenocarcinoma 1 (202) 1 (60) 5 (446) TCGA rectum adenocarcinoma 2 (178) 2 (144) 2 (334) TCGA skin cutaneous melanoma 1 (242) 1 (154) 2 (442) TCGA stomach adenocarcinoma 2 (212) 1 (58) 5 (696) TCGA thyroid carcinoma 1 (260) 1 (86) 5 (715) TCGA uterine corpus endometrioid carcinoma 2 (488) 1 (53) 3 (448) 1 (200) 6 (2312) SU2C Breast Public 1 (54) 2 (92) CCLE 2 (1934) 1 (972) Other datasets from the literature 19 (3556) 17 (2206) Number of datasets by cancer type and data type; values in parenthesis are number of samples. most of which are from The Cancer Genome Atlas project NEW DATA (TCGA, https://tcga-data.nci.nih.gov/tcga/) (5). Data on The Cancer Genome Atlas the website are updated periodically to include the latest We provide an open-access portal to view, analyse and releases from TCGA and other projects. Currently, the download public data from TCGA. We have developed browser holds 201 TCGA public-tier datasets from 22 an automated pipeline that loads and processes data TCGA cancer projects, data from the Cancer Cell Line from TCGA, allowing it to be quickly downloaded from Encyclopedia project (CCLE, http://www.broadinstitute. their servers and then displayed in the browser. This org/ccle/home) (6), and 43 other published studies. The pipeline has greatly increased the number of datasets Google-maps-based microscope slide viewer has 2433 from TCGA we can load into the browser, which has slides from TCGA. allowed us to expand the cancers we have available to A controlled access mechanism is also available for 22 cancer projects including breast, pancreas and lung private data, and currently supports the Stand Up To cancer (Table 1). Cancer breast cancer dream team (SU2C, http://www. New dataset types include GISTIC2 (9) estimated seg- standup2cancer.org/dream_teams/view/an_integrated_ mented copy number variation from the TCGA approach_to_targeting_breast_cancer_molecular_ FIREHOSE pipeline. We display next-generation RNA subtypes_and_th), I-SPY 2 TRIAL (Investigation of Illumina HiSeq gene expression data, protein expression Serial Studies to Predict Your Therapeutic Response data assayed by reverse phase protein array technology, with Imaging And moLecular Analysis 2) (7), LINCS and DNA methylation profiles measured using the (Library of Integrated Network-based Cellular Signatures) (http://www.broadinstitute.org/LINCS/) (8) Illumina Infinium HumanMethylation450 platform. We and other projects. The controlled access mechanism also have datasets showing integrated gene activity level allows authorized users to view this private data alongside inferred using the PARADIGMLITE method. public data. PARADIGM (10) is pathway analysis method to infer In the past 2 years, we have made significant changes patient- or sample-specific genetic activities by incorporat- to the browser including improving the user interface, ing curated pathway interactions with genomic measure- as well as implementing user accounts, an online tutorial ments, such as gene expression and copy number data. and better documentation. The data versioning system PARADIGMLITE, a less computationally intensive we developed this year, allows us to offer users the version of PARADIGM, integrates different types of ability to download datasets. We also released a new data on individual genes without incorporating the Tumor Image Browser, based on Google Maps, that curated pathway interactions. The gene expression and offers intuitive panning and zooming across microscope copy number input data are shown in the accompanying slide images. evidence datasets. Nucleic Acids Research, 2013, Vol. 41, Database issue D951 Figure 1. TCGA GBM datasets showing differential copy number variation, gene expression and methylation for the glioma-CpG island methylator phenotype (G-CIMP). The black box emphasizes the samples characterized as G-CIMP tumors. Copy number and DNA methylation datasets, by default, use red and blue to represent amplification and deletion, respectively. Gene expression datasets, by default, use red and green to represent over- and under-expression, respectively. For the G-CIMP clinical feature, yellow represents tumors characterized as G-CIMP and black represents tumors who are not. For the gene expression subtype clinical feature the four subtypes, from black to bright yellow, are proneural, neural, classical, and mesenchymal. (A) TCGA GBM whole genome copy number profile. (B) TCGA GBM gene expression for a select set of genes. (C) TCGA GBM DNA methylation for a select set of genes. The clinical data, which is displayed alongside the screenshot of some TCGA Glioblastoma multiforme genomic information, have also been updated with more (GBM) datasets showing differential copy number vari- readable clinical feature names and values. This allows for ation, gene expression and methylation for the easier access for users who may not be familiar with the glioma-CpG island methylator phenotype (G-CIMP). abbreviations used by TCGA. Figure 1 illustrates some of Viewing the whole genome copy number profile, users these more readable clinical feature names as well as a few can easily see that the G-CIMP tumors display distinct of the new data types from TCGA. Figure 1 is a browser copy number alterations, lacking the deletion of D952 Nucleic Acids Research, 2013, Vol. 41, Database issue chromosome 10 and amplification of chromosome 7 making the controls more obvious and intuitive, enhancing shared by the majority of the non-G-CIMP samples. the readability of drawn text, and improving the responsive- The browser also shows that G-CIMP samples are ness of the site by using more modern web technologies. highly enriched in the proneural subtype, as described in Dataset selection has been moved closer to the top of the recent publication that identifies G-CIMP samples as a the page, making it easier for users to find datasets of distinct subset of human gliomas on molecular and clinical interest. All per-dataset controls have been moved to each grounds (11). When viewing genes linked to survival individual dataset display, allowing direct configuration. outcome in GBM (12) we see that individuals with Additionally, it is now easier and more natural to find or G-CIMP tumors have different gene expression and make user genesets and signatures. The user interface for the DNA methylation profiles compared with non-G-CIMP clinical heatmap has been dramatically improved allow- tumors, and a higher survival rate after treatment, as ing intuitive configuration of visible clinical features and described in the recent publication (11). subgroup construction based on clinical and genomic data. In tandem with the updated interface, the browser now Cancer Cell Line Encyclopedia has an interactive tutorial which highlights much of the new functionality. Our software detects when a user last Our newest public data are from the CCLE, a reference used the browser and automatically opens if new features library containing primarily genetic and pharmacologic have been added to the browser since they last visited. information from a large panel of human cancer cell line Additionally, the user guide was completely updated and models (6). The CCLE datasets includes genome-wide FAQs were added. copy number and gene expression profile of the cell lines, mutations assessed in 33 genes, and pharmacological response to 24 anti-cancer drugs. These datasets are avail- New clinical heatmap features able to the public. In addition to updating the interface for the clinical heatmap, some new features were added. It is now Stand Up To Cancer public datasets from the Gray lab possible to zoom in on a subset of samples by clicking We are also displaying published copy number profile and and dragging within the clinical heatmap. This makes it gene expression levels of a collection of breast cancer cell easier to more closely examine a sample and its values. lines from Joe Gray’s lab at Oregon Health and Science Clinical features are also now automatically sorted from University in association with SU2C (13,14). Response to left to right. To modify the sort order of the features in the 77 therapeutic compounds (GI50 scores), ER and HER2 heatmap, users click to move clinical features. Since receptor status, and other phenotypic information is sample order in the genomic and associated clinical shown in the clinical data. heatmap is determined by the sort order of the clinical features, rearranging clinical features automatically triggers a vertical reordering of both heatmaps. This NEW FEATURES allows users to more easily explore phenotype and Website redesign genotype associations using the browser. Additionally, The largest new feature is a completely redesigned user subgroups are now displayed as a clinical feature, interface (Figure 2). In particular, we have focused on making it easier to sort samples by any subgrouping. Figure 2. Screenshot from user documentation highlighting features in the new user interface. Nucleic Acids Research, 2013, Vol. 41, Database issue D953 Updated genes view from ovarian serous cystadenocarcinoma, 1273 from glio- blastoma multiforme and 306 from uterine corpus Users can view genomic data on a per-gene basis, where endometrioid carcinoma. probes that map to the same gene are displayed together. To display DNA methylation data from platforms with many probes per gene, especially the Infinium FUTURE DIRECTIONS HumanMethylation450 chip, we now render probes Our plans for the next year include providing users a way ordered left to right in the direction of transcription. to bookmark a view both for themselves and to share with Used in conjunction with chromosome view, this allows others. Internal progress has already been made allowing the user to easily see which section of a gene is methylated users to visualise their own patient-identifiable or unpub- under genes view, allowing evaluation of DNA methyla- lished data in a secure manner. We will also develop new tion in different genomic contexts, such as multiple viewing capabilities for multianalyte data, allowing user to promoter sites and in gene bodies (Supplementary view copy number, gene expression, and clinical data from Figure S1). The relationship between DNA methylation the same set of samples within the same map. Finally, we and transcription is more nuanced than first predicted, are planning on displaying mutation data in a new making it important to visualize exactly where in a gene heatmap visualization. the methylation is occurring (15–17). We will continue to expand the data we display from TCGA. Additionally, we will display the new public data User accounts and website architecture from LINCS, which aims to use genome-wide expression Previously, there were many browser installations at dif- profiling to catalog the cellular consequences of small- ferent websites for groups desiring private access to pro- molecule and genetic perturbations in a breadth of tected data. To reduce engineering overhead, a controlled human cell lines. We plan to publicize and provide an access mechanism was put in place to restrict the display API for our data format. of certain data to authorized users. This mechanism relies on new backend architecture that allows us to provide the same level of security as private installations but is SUPPLEMENTARY DATA transparent to the user. As part of this, we implemented Supplementary Data are available at NAR Online: user accounts, allowing us to authenticate users. This new Supplementary Figure 1. security system currently supports the SU2C breast cancer dream team, ISPY 2 TRIAL, LINCS project, and other projects. ACKNOWLEDGEMENTS Implementing user accounts resulted in two additional The authors thank Erich Weiler, Jorge Garcia and the features. First, users who had access to multiple private UCSC Genome Browser for their support. installations can now view data from all of their projects together. This allows users to make connections across data sources. User accounts also allowed us to start FUNDING saving user genesets and signatures. As we offer more National Cancer Institute (NCI) [5U24CA143858]; new features, we hope to allow users to save other data American Association for Cancer Research (AACR) as well, including bookmarks. Users do not need to be Stand Up To Cancer (SU2C) program [SU2C-AACR- authorized to sign up for a free account or access public DT0409]; UCSF Helen Diller Family Comprehensive data; this service is open to everyone. Cancer Center. Funding for open access charge: Howard Hughes Medical Institute. Data version control and downloads When we constructed the TCGA pipeline, we concur- Conflict of interest statement. None declared. rently implemented data freezes and data version control. This was done so users know what version of REFERENCES the data they are viewing and whether the data changed since it was viewed last. Users can also download the pro- 1. Hahn,W.C. and Weinberg,R.A. (2002) Modelling the molecular cessed datasets for further analysis. These processed circuitry of cancer. Nat. Rev. Cancer, 2, 331–341. 2. Hanahan,D. and Weinberg,R.A. (2011) Hallmarks of cancer: the datasets can have advantages over the original data as next generation. Cell, 144, 646–674. they have been normalized and include clinical data 3. Vogelstein,B. and Kinzler,K.W. (2004) Cancer genes and the curation. pathways they control. Nat. Med., 10, 789–799. 4. Zhu,J., Sanborn,J.Z., Benz,S., Szeto,C., Hsu,F., Kuhn,R.M., Karolchik,D., Archie,J., Lenburg,M.E., Esserman,L.J. et al. Tumor Image Browser (2009) The UCSC Cancer Genomics Browser. Nat. Methods, 6, Our previous Tumor Image Browser, hgMicroscope, has 239–240. 5. Collins,F.S. and Barker,A.D. (2007) Mapping the cancer genome. been replaced by a completely new Google-Maps based Pinpointing the genes involved in cancer will help chart a new browser. Applying the Google Maps interface to tumor course across the complex landscape of human malignancies. Sci. images has made exploring a microscope slide easier and Am., 296, 50–57. more intuitive. Currently the browser contains 2748 6. Barretina,J., Caponigro,G., Stransky,N., Venkatesan,K., TCGA images, both tumor and normal, including 1169 Margolin,A.A., Kim,S., Wilson,C.J., Leha´ r,J., Kryukov,G.V., D954 Nucleic Acids Research, 2013, Vol. 41, Database issue Sonkin,D. et al. (2012) The cancer cell line encyclopedia enables phenotype that defines a distinct subgroup of glioma. Cancer predictive modelling of anticancer drug sensitivity. Nature, 483, Cell, 17, 510–522. 603–607. 12. Colman,H., Zhang,L., Sulman,E.P., McDonald,J.M., 7. Barker,A.D., Sigman,C.C., Kelloff,G.J., Hylton,N.M., Berry,D.A. Shooshtari,N.L., Rivera,A., Popoff,S., Nutt,C.L., Louis,D.N., and Esserman,L.J. (2009) I-SPY 2: an adaptive breast cancer trial Cairncross,J.G. et al. (2010) A multigene predictor of outcome in design in the setting of neoadjuvant chemotherapy. Clin. glioblastoma. Neuro Oncol., 12, 49–57. Pharmacol. Ther., 86, 97–100. 13. Heiser,L.M., Sadanandam,A., Kuo,W.L., Benz,S.C., 8. Garnett,M.J., Edelman,E.J., Heidorn,S.J., Greenman,C.D., Goldstein,T.C., Ng,S., Gibb,W.J., Wang,N.J., Ziyad,S., Tong,F. Dastur,A., Lau,K.W., Greninger,P., Thompson,I.R., Luo,X., et al. (2012) Subtype and pathway specific responses to anticancer Soares,J. et al. (2012) Systematic identification of genomic compounds in breast cancer. Proc. Natl Acad. Sci. USA, 109, markers of drug sensitivity in cancer cells. Nature, 483, 570–575. 2724–2729. 9. Mermel,C.H., Schumacher,S.E., Hill,B., Meyerson,M.L., 14. Neve,R.M., Chin,K., Fridlyand,J., Yeh,J., Baehner,F.L., Fevr,T., Beroukhim,R. and Getz,G. (2011) GISTIC2.0 facilitates sensitive Clark,L., Bayani,N., Coppe,J.P., Tong,F. et al. (2006) and confident localization of the targets of focal somatic A collection of breast cancer cell lines for the study of copy-number alteration in human cancers. Genome Biol., 12, R41. functionally distinct cancer subtypes. Cancer Cell, 10, 515–527. 10. Vaske,C.J., Benz,S.C., Sanborn,J.Z., Earl,D., Szeto,C., Zhu,J., 15. Curradi,M., Izzo,A., Badaracco,G. and Landsberger,N. (2002) Haussler,D. and Stuart,J.M. (2010) Inference of patient-specific Molecular mechanisms of gene silencing mediated by DNA pathway activities from multi-dimensional cancer genomics data methylation. Mol. Cell. Biol., 22, 3157–3173. using PARADIGM. Bioinformatics, 26, i237–i245. 16. Eden,S. and Cedar,H. (1994) Role of DNA methylation in the 11. Noushmehr,H., Weisenberger,D.J., Diefes,K., Phillips,H.S., regulation of transcription. Curr. Opin. Genet. Dev., 4, 255–259. Pujara,K., Berman,B.P., Pan,F., Pelloski,C.E., Sulman,E.P., 17. Jones,P.A. (2012) Functions of DNA methylation: islands, start Bhat,K.P. et al. (2010) Identification of a CpG island methylator sites, gene bodies and beyond. Nat. Rev. Genet., 13, 484–492.

Journal

Nucleic Acids Research – Oxford University Press

Published: Jan 29, 2013

There are no references for this article.