1. Investigating structural variants

Not every cancer has determining somatic mutations. Using the full power of WGS data, relevant structural variants can be traced also and linked to potential causes of disease

In this course we’ll introduce R2, the web based genomics analysis and visualization application. Throughout the course an integrative approach to genomics data will be used. By combining sequencing data with expression data and vice versa new insights can be derived. Throughout this course we’ll focus on data of the childhood tumor neuroblastoma.

We hope to show how R2 can be used to visualize and analyze your WGS data.

This resource is located online at http://r2-training-courses.readthedocs.io

1.1. Introduction

Cancer is a very complex disease. Much more complicated than originally anticipated when the first mutations were found to be causal for specific cancers. At that time, for colorectal cancer, a well defined path of subsequently gained mutations was found to lead to more aggressive tumorigenic cell types (the Vogelstein model).

Figure 1: Mutation paths during cancer progression

Figure 1: Mutation paths during cancer progression.

Although there has been extensive research into similar mutation mechanisms in neuroblastoma (also in the AMC Oncogenomics group), such a mechanism has not been found for this type of cancer. In this practical work session we will try to bring you cutting edge research in this often deadly childhood tumor.

Recent research suggests that neuroblastoma consists of different cancer cell types. There is reason to believe that this heterogeneity causes the high percentage of relapses in the aggressive subtype of neuroblastoma. Children developing a relapse almost always die. Fortunately new technologies have become available to molecular biology. These enable us to not only study mutations and RNA expression of genes, but to study the epigenetic modifications of the DNA-associated histones as well. And in addition, genes can now be manipulated in cell lines and living tissues.Using advanced data analysis, statistics and clustering methods, the field of bioinformatics tries to derive new insights from these experimental data and help molecular biologists to generate hypotheses that can be tested experimentally. Today you will use the web-based genomics analysis and visualization platform R2. R2 provides you with a set of bioinformatics tools to investigate recent patient and experimental data from neuroblastoma tumors and cell lines.

Despite decades of research high stage neuroblastoma still has a very poor prognosis. Since cancer is a disease of genomic aberrations we’re first going to investigate what aberrations are present and how these might relate to the onset of neuroblastoma.

1.2. Exploring the dataset

The oncogenomics department of the AMC has gathered a richly annotated set of neuroblastoma tumors. To easily explore this the R2 development team has devised the concept of Datascopes; a convenient view on the data with some pre-built analyses readily available.

  • Go to R2 (http://r2.amc.nl) by clicking on the button below:
  • Log on to the R2 platform with your credentials that were provided. (or apply for a login using the link)
  • In the left menu click on Change Data Scope > Training > Graduate Training Course
  • In the middle section of the page, an additional choice step appears; click Goto Graduate Training Course home
  • For a quick impression of the data select the Cohort Overview R2 presents the tumor series with it’s annotation. Explore the distribution of the parameters.

_images/R2d2_logo.pngClick on the distributions of different parameters. How many samples have a MYCN amplification?




Until recently only several genomic aberrations were known:

Gene Type Reference
MYCNamplification 20%(Schwab et al., 1983)
ALK7%(George/Mosse/Janoueix-Lerosey/Chen 2008)
CyclinD1amplification 4%(Molenaar et al., 2003)
PHOX2B4%(van Limpt et al., 2004)
PTPRD4%(Stallings et al. 2006)
NF13%(Hölzel et al., 2010)
PTPN112%(Merks et al., 2004, Bentires-Alj et al., 2004)
FOXR11%(Santo et al., 2011)
LIN28B1%(Molenaar et al., 2012)

To extend these data, the Oncogenomics department of the AMC set out to sequence 87 untreated primary neuroblastoma tumours of all stages from this set.

1.3. Somatic mutations in neuroblastoma

For this the samples where sent to the Complete Genomics sequencing facility, now taken over by BGI, for whole-genome paired-end sequencing. They provide a sequence as a service model. Genomes were sequenced at an average coverage of 50. Compared to the HG18 reference genome an average of 3,347,592 singlenucleotide variants (SNVs) per genome were obtained, in accordance with reported frequencies of interpersonal variants.

The R2 development team has processed these WGS data further using the CGAtools software to compare tumor with lymphocyte genomes. This provided a somatic score, estimating the likelihood of mutations to be somatic. Through several filtering steps the somatic mutations were determined with respect to the reference genome.

Figure 2: Comparing Tumor data with the reference genome from Lymphocytes.

Figure 2: Comparing Tumor data with a reference genome from Lymphocytes.

A comprehensive list of the mutations can be accessed through R2.

  • Go back to the Graduate Training Course datascope (still open in another tab)
  • Select the somatic variants tile
  • A table with all mutations in the 86 tumors appears in a new tab. It is basically a view on a database table. Ordering on its columns is possible by clicking on the column header. Sort the column by gene name.

_images/R2d2_logo.pngCan you spot recurring mutations?




  • There are no mutations recurring more than a few times. Go to the ALK gene and select the view link (note: this is separate from the detail link). In a new tab this mutation is shown in the R2 genomebrowser zoomed in on the genome to the base level. All samples are drawn beneath this stretch. Annotation of the publicly available cosmic mutation database is provided as well.

_images/R2d2_logo.pngWhat type of aberrations does the ALK gene suffer?




  • The buttons on top of the page can be used to zoom in and out. Zoom out 4 times with a factor 20.
  • The GenomeBrowser has a tremendous number of parameters that can be set. Scroll down to the lower half of the page. A form shows quite some parameter fields. These provide additional annotations and settings for the algorithms used. A useful annotation is provided by the NIH epigenome roadmap that annotates the genome with chromatin modification data, which is based on methylation and acetylation patterns of the genome. This annotation however, is only provided on another Human Genome build. In the Adjustable settings form change the GenomeBuild to HG19 (note that other builds as well as mouse data is available also). Click redraw
  • An unannotated version of the reference genome is shown. Find the Refseq(R2) and switch the annotation on. Click redraw

_images/R2d2_logo.pngWhat has happened to the ALK gene?




  • If ALK gene is out of scope, locate the ALK gene (hint type in the gene name in the left upper Find gene field)

  • For clarity you can now switch off the cosmic annotation (in the Genome Variation box) and the Calldif Somatic Genome annotation (in the X:Complete Genomics => Variants box). Set the NIH Epigenome Roadmap annotation to all (in the TranscriptView annotation box). This annotation provides information on public datasets that have established whether chromatin regions are subject to active transcription (green), enhancer regions (yellow) or are part of a transcription start site (red).

  • Select the front end of the gene by selecting a region; see image (hint the color of the transcript denotes the reading direction; green means the regular direction, red the opposite direction) Click redraw (Note: the NIH annotation only appears for regions under 200.000 bp)

    Figure 3: Selecting a region.

    Figure 3: Selecting a region.

  • This NIH Epigenome Roadmap annotation is actually a sum of data from a lot of data sources. These sources can be further detailed by selecting the detail view in the toolbox that appears when you click the tools icon, see image below. This box appears at more settings fields if available.

    Figure 4: Opening the NIH parameter settings toolbox.

    Figure 4: Opening the NIH parameter settings toolbox.

    Figure 5: Annotation colors with the chromatin state description.

    Figure 5: Annotation colors with the chromatin state description.


_images/R2d2_logo.png**What chromatin annotation is available for the start of the ALK gene? (You can hover over the annotation bars to see the ) **




  • Now go back to the Graduate Training Course datascope, select the Somatic Variants Table tile and now click the detail link. R2 shows additional information on the expression of the gene and its location on the genome.

_images/R2d2_logo.pngWhat is remarkable about the expression of the ALK gene?




  • From this detail view other analysis tools within R2 can be approached by clicking on the links below the graphs. Feel free to explore these further.

1.4. Further use of WGS data; structural variants

WGS data allows for further analysis; the paired end technique enables the discovery of structural variants.

Figure 6: Paired end sequencing makes discovery of structural variants possible.

Figure 6: Paired end sequencing makes discovery of structural variants possible.

  • These structural variations are best visualized as so called circosplots. To access these in R2 go to the Graduate Training Course datascope and click the circos archive tile.
  • An overview of all sequences appears displayed as circos plots. These give an immediate comprensive view on the state of the genome. Click on one of the circos plots.
  • In a new tab a detailed view of this specific tumor genome is shown. When hovering over the plot the mouse opens a magnifier window.

_images/R2d2_logo.pngWhat do the green and red areas mean? And the lines crossing the circle?




  • In the tabbed panel to the right of the circos plot all information is detailed. Open the sample annotation tab.

_images/R2d2_logo.pngWhat are the patient characteristics?




  • Now open the Somatic Structural Variants panel.

_images/R2d2_logo.pngCan you locate a structural variant that involves a gene and spans two chromosomes? (Note: clicking on the view link shows the actual locations on the genome)




1.5. Chromothripsis

While investigating the WGS data, an interesting phenomenon was observed. In some tumor samples parts of the genome appeared to be riddled with structural variants, resulting in a shredded chromosomal structure.

  • Go to the overview page with circos plots.

_images/R2d2_logo.pngCan you spot an example of chromothripsis from the circos overview?




To see how chromothripsis relates to clinical data we can investigate survival data in R2.

  • From the left menu in the main Graduate Training Course datascope panel select Kaplan Meier > By annotated parameter
  • Make sure that the appropriate neuroblastoma set has been selected in the dataselection panel: ‘Tumor Neuroblastoma (combat) - Versteeg - 122 - MAS5.0(bc) - u133p2’. Click Next.
  • A selection menu appears, in the use track field select the track cg_chromothripsis and click next

_images/R2d2_logo.pngHow does chromothripsis affect survival?




  • Filtering for subsets allows you to further isolate specific survival characteristics. If you like you can toy around with different parameters.

1.6. Locations of structural variants, hotspots?

Chromothripsis can be seen as an extreme case of concentration of structural variants in one sample. The question arises whether there are other hotspots of structural variation on the genome that are found in multiple samples. These might point to functional interactions.

  • One of the genes that exhibited such a hotspot is the TERT gene. Go back to the startpage of the Graduate Training Course datascope and select the GenomeBrowser tile. This brings you to the TERT gene on the genome with some preset annotations.

_images/R2d2_logo.pngHow does this region qualify as a hotspot?




  • As is obvious from the high number of peaks around the TERT gene there is a hotspot of structural variants in that area. The types of variants are annotated below the stretch on the genome.

_images/R2d2_logo.pngWhat do the arrows and colored tracks mean? (Hint: hovering with the mouse provides additional information)




  • Red arrows depict translocations to other loactions in the genome. Locate the translocation to chromosome 11 for sample 724 (Hint: hovering over the arrows gives sample information) and click on the arrow.
  • R2 brings you to the other side of the translocation. In the TranscriptView panel switch on the SuperEnhancers NB annotation and click redraw.

_images/R2d2_logo.pngWhat advantage would a tumor have from this translocation?




  • To further corroborate this we can go to the Circos plots panel again. Go back to the Graduate Training Course datascope overview panel and click the circos archive tile again.
  • Locate the N724 tumor sample and click on the image.
  • Open the Gene Expression list tab
  • This can be further explored by clicking on the probeset link (left column in the list) and on the Detailed link.

_images/R2d2_logo.pngHow is the expression of the TERT gene affected?




1.7. Remarks

This ends the first part of this course. You can continue now with the analysis of intra-tumor heterogeneity. We hope that this course has been helpful. At the end of the workshop, please provide feedback on the course with this form.

If you want to have your genomics data visualized and analyzed using the R2 platform you can always consult r2-support@amc.nl

The R2 support team.