Whole genome sequencing: What additional results do we get?

In this step, we will explore considerations around the results returned to genomics laboratories from samples sent for whole genome sequencing.

Although whole genome sequencing (WGS) is not yet a standard of care test in the NHS, it is important to recognise and understand the different types of data that might appear exclusively on a WGS report. Below, we’ll explore some, and think about they might be interpreted and used in practice.

Introducing the types of data

The key difference between standard of care genomic testing results and results returned from WGS is the global or genome-wide nature of WGS analysis and the complexity of the information received. Results are returned to clinicians in a form that is approachable and that can be applied practically for a given patient. The data returned from WGS can be split into the following types:

• circos plots;
• small somatic variants;
• tumour mutational burden;
• mutational signature analysis;
• rainfall plots; and
• constitutional (germline) data.

Remember: All of these results need to be triaged and analysed to identify any potential oncogenic drivers or tumourigenic variants or any other changes of therapeutic significance.

Note: Images in this step and the following step were provided courtesy of Genomics England with permission for educational purposes.

Circos plots

Circos plots are a genome-wide summary of the data that aims to incorporate single nucleotide variant, copy number variant and structural variant data to illustrate the overall picture of the genetic complexity of a tumour. Below is an example labelled with what each section is telling us.

_{Figure 1: Circos plot labelled with the chromosomes, structural rearrangements, depth of coverage, copy number changes and the number of somatic indels and single nucleotide variants (SNVs).}

Working from the inside-out, the lines in the centre of the plot represent structural rearrangements. The coloured bars these lines connect represent individual chromosomes and all the data is mapped against these. The red dots immediately outside of the chromosomes are the number of somatic single nucleotide variants, or SNVs, in that region. The green line directly outside of those is the number of somatic indels, or insertions and deletions, in that region. And then the next ring out covers copy number changes, with losses in red, gains in green, and copy number neutral loss of heterozygosity, which we’ll talk about a bit more later, in orange. Finally, the outermost ring represents depth of coverage.

Circos plots: Examples

_{Figure 2: Example circus plots. Plot A is relatively simple with a small number of structural rearrangements and single nucleotide variants (SNVs). Plot B shows a highly complex genome with a large number of structural rearrangements and copy number changes.}

Now, let’s look at two examples (figure 2). You can see from these two circos plots that the data can look very different depending on the type of tumour. Plot A, a genome from a paediatric Wilms tumour, is less complex. You can see that there are a small number of structural rearrangements and single nucleotide variants, as indicated by the lines in the centre and the red dots immediately outside of the coloured chromosome segments, respectively.

Plot B is from an adult spindle cell sarcoma and is a highly complex genome with a large number of structural rearrangements. This is clear both from the lines in the middle and from the copy number changes around the outside.

Small somatic variants

The term small somatic variant covers both single nucleotide variants, or SNVs, and small indels. Most samples have a large number of these. As an example, in the last year at the central lab of the North East and Yorkshire Genomic Laboratory Hub (GLH), WGS results returned between 30 and 182 small somatic variants. Each individual sample had an average of 54 such variants, but the exact number detected is highly dependent on tumour type. It’s impossible to analyse this many variants for this many samples within at the NHS at present, and so they have to be prioritised based on tumour type and the nature of the variant.

_{Figure 3: The filtering used to prioritise small somatic variants for analysis.}

This data is returned to the laboratories already filtered into three domains, 1, 2, and 3:

• Domain 1: In general, includes the genes and variants most likely to be potentially clinically actionable for the patient. Contains variants in genes that are included in the National Genomic Test Directory for cancer.
• Domain 2: Includes variant types and genes and may be actionable, and are more significant for the tumour type. Contains variants in genes associated with the specific patient’s tumour type.
• Domain 3: Everything else on the cancer census. Note: only variants in domains 1 and 2 are currently analysed.

Using the example from the North East and Yorkshire Lab, this reduces the number of variants to be analysed for each patient from an average of 54 to an average of three, which is far more manageable.

Tumour mutational burden

The tumour mutational burden is a more global measure of what is going on within a tumour. Mutational burden can be understood as the number of somatic, non-synonymous, small variants per megabase of coding DNA. On the graph in figure 4 below, each dot represents an individual tumour. The tumour mutational burden for a given tumour is on the y-axis and the tumour type is on the x-axis. The tumour mutational burden of the sample being analysed is given as a dotted red line across the middle for comparison to the expected number for any given tumour type.

_{Figure 4: Tumour mutational burden of a range of tumour types.}

So, how can this help? A high tumour mutational burden – one greater than, say, 10 – would be consistent with a defect in the mismatch repair pathway. On the other hand, a very high tumour mutational burden – greater than 100 – would be consistent with loss of function of the proofreading genes, for example, POLE, so this data can be used to inform analysis of any variants found within a tumour.

Rainfall plots

Rainfall plots are a genome-wide cluster analysis of small somatic variants. The aim of this visualisation is to examine the spatial distribution of small somatic variants across the genome. This allows us to detect both small areas of mutation, along with larger clusters known as kataegis. These are key markers of genomic instability and associated with loss of function of the APOBEC family of single-stranded DNA repair enzymes. Each dot on the rainfall plot represents a small somatic variant within the tumour being looked at. The x-axis plots the chromosome position of this variant and the y-axis plots its distance along the chromosome relative to the previous small somatic variant.

_{Figure 5: An example rainfall plot.}

A rainfall plot allows us to look for regions of hypermutation within the tumour, and these are marked on the graph (figure 5) by the small black circled arrows at the bottom. Hypermutation and kataegis are most often seen in breast cancers, but can present in other tumour types.

Mutational signature analysis

Mutational signatures are characteristic patterns of somatic mutations in cancer genomes reflecting the underlying mutational processes. We can think of them as the molecular footprint of a tumour. These signatures have been identified by very large-scale bioinformatic analysis of sequencing data sets, and we have identified 30 different signatures up to this point. Any given tumour can contain multiple different mutational signatures, and the relative contributions of these are defined mathematically. You can see the contributions of different signatures in figure 6, given as percentages. At the current time mutational signatures are mainly used in research, but they may be used in far more detail in clinical settings in the future. If you’d like to learn more about mutational signatures, view the related page on GeNotes.

_{Figure 6: The percentage contributions of current identified mutational signatures.}

Constitutional (germline) data

For WGS analysis, both somatic (tumour) and constitutional (germline) samples are provided, and this allows acquired somatic variants to be distinguished from constitutional ones. Additionally, it allows us to identify incidental findings within the germline and raises the question of what constitutional data should be analysed and reported. Currently, only those changes directly relevant to the tumour type being analysed are looked at, plus a small list of variants and genes considered to be highly significant. These would include, for example, BRCA1 and BRCA2. It should be stressed that this is not a substitute for constitutional-specific WGS as the coverage isn’t as good and it is not as informative as a trio analysis.

_{Figure 7: The filtering used to prioritise constitutional (germline) variants for analysis.}

The constitutional data that is returned is filtered into two groups, tier 1 and tier 3 (figure 7). In tier 1, there are potentially pathogenic variants in cancer susceptibility genes relevant to the reported disease type, and we analyse these according to the guidance.

Tier 3 contains all variants across a large panel of cancer susceptibility genes. A small number of these genes are considered highly significant and are always analysed, such as BRCA1 and BRCA2 that we mentioned earlier, but other genes in this tier would only be analysed if the clinician reports that there is a high likelihood of inherited cancer syndrome for the patient or their family. All other constitutional variants are not currently returned to the lab for analysis. Please note that this represents policy at the time of writing, but be aware that it is subject to change.

Next, it’s time for a quick quiz to consolidate your knowledge of what you’ve learned about results returned from WGS.