Confirming identities of replicate & matched samples with SNP targeted assays in methylation arrays

Methylation studies frequently include samples derived from multiple tissues from the same individual (matched samples). In addition, technical or biological replicates may be included to assess or confirm data quality and reproducibility. The ability to confirm the genotypic identity of replicate and matched samples is important when sample mapping is needed.

Infinium Methylation BeadChips include a set of probes targeting single nucleotide polymorphisms (RS-SNPs), rather than CpG methylation sites. Specifically, the Infinium MethylationEPIC v1 array has 59 RS-SNP assays, while both the Infinium HumanMethylation450 and the Infinium MethylationEPIC v2 BeadChip include 65 RS-SNP assays. These SNP-based probes can be used to verify the identity of replicate and matched samples derived from the same individual. This identification is based on the fact that these SNP-based assays target genetic variation rather than methylation, they behave similarly to genotyping assays. For a given SNP locus:* Samples with an AA genotype will typically produce beta values clustering near 0.0

• Samples with an AB genotype will cluster around 0.5

• Samples with a BB genotype will cluster near 1.0 Thus, samples derived from the same individual are expected to display highly concordant beta values across the spectrum of RS-SNP probes. Conversely, samples derived from different individuals will exhibit discordant and more widely scattered beta value distribution.

The following methods assume familiarity with GenomeStudio, including the ability to:

• Generate scatter plots

• Perform cluster analysis and create dendrograms

• (Optionally) Use filtering tools to select a subset of the entries in a table Both methods described below rely exclusively on RS-SNP probe data and are applicable to Infinium MethylationEPIC v1, Infinium HumanMethylation450 and the Infinium MethylationEPIC v2 datasets.

Method 1:

Scatter Plot analysis to determine whether two selected samples are derived from the same individual

This method involves generating scatter plots to directly compare RS-SNP beta values between two selected samples. It uses only the RS-SNP assays present on the Infinium MethylationEPIC BeadChip (and is also applicable to Infinium HumanMethylation450 and Infinium MethylationEPIC v2 data) to assess one pair of samples at a time, and determine whether the samples are derived from the same individual.

1. In the Sample Methylation Profile table of the GenomeStudio 2011 Methylation project, scroll to the bottom and select all the TargetIDs that start with “rs”.

• Right-click anywhere within the table, and from the context menu, select “Show only selected rows.” The table should now display only 65 (or 59) rows, all corresponding to TargetIDs that begin with “rs”.

• Alternatively, open the Filter tool and apply the condition “TargetID has rs”.

  

Figure 1. Filter functionality in GenomeStudio v2011.1 within the Sample Methylation Profile table.

Figure 2. Use of the filter function in GenomeStudio v2011.1 to select RS‑SNP assays based on TargetID criteria.

Figure 3. Sample Methylation Profile table in GenomeStudio v2011.1 displaying RS‑SNP probes after filtering by TargetID (“rs”).

1. Select the Scatter Plot icon located in the toolbar above the Sample Methylation Profile table. A dialog box titled “Scatter Plot” will appear.

Figure 4. Scatter Plot configuration window in GenomeStudio v2011.1, showing selection of sample columns and AVG_Beta values for pairwise comparison.

1. In the Columns list, select one of the two samples of interest. Ensure that the sub column is set to “AVG_Beta”, then select the “Y Axis” button.

2. In the Columns list, select the second sample. Again, ensure that the sub column set to “AVG_Beta", then select the “X Axis” button.

3. Select “OK” to generate the scatter plot.

4. If the two samples are derived from the same individual, three distinct clusters of data points will be observed, one in the lower-left quadrant of the plot, one in the center, and one in the upper-right, reflecting the expected genotype groupings.

   

Figure 5. Scatter plot of RS‑SNP beta values (AVG_Beta) comparing two samples derived from the same individual, showing tight clustering along the diagonal and distinct genotype groupings.

1. If the two samples are derived from different individuals, nine distinct clusters of datapoints will typically be observed, arranged as three clusters across each of the lower, middle, and upper regions of the plot.

   

Figure 6. Scatter plot of RS‑SNP beta values (AVG_Beta) comparing two samples derived from different individuals, showing dispersed data points and lack of alignment along the diagonal. Method 2:

Using Cluster Analysis to identify samples that are likely derived from the same individual

This method generates a dendrogram based on the RS-SNP beta values to evaluate similarity across multiple samples simultaneously. By visualising sample relationships in a dendrogram, users can readily identify clusters of samples that are highly correlated and therefore likely to originate from the same individual.

1. Follow Steps 1 and 2 from Method 1 to select the RS‑SNP probes and open the relevant workspace.

2. Select the “Sample Methylation Profile” tab to ensure it is the active table. The selected tab should be highlighted in light blue (see Figure 7).

3. In the GenomeStudio menu bar, select the “Run Cluster Analysis” icon, depicted as a miniature dendrogram. A dialog box titled “Cluster Analysis: Sample Methylation Profile” will appear.

Figure 7. GenomeStudio v2011.1 interface showing the Run Cluster Analysis option within a methylation project.

1. To include all samples in the analysis, select the “Select all” button at the bottom of the window. Alternatively, to analyze a subset of samples, select the desired samples from the list under the “Groups” section.

Figure 8. Cluster Analysis dialog window in GenomeStudio v2011.1, illustrating sample selection and configuration of clustering parameters (Samples and correlation metric).

1. In the “Cluster” section, select “Samples”. Under the “Metric” option, choose the correlation method to be used for the analysis (see Figure 8).

2. Select “Create Dendrogram” to generate the clustering results.

3. Samples derived from the same individual are expected to show extremely high correlation, typically greater than 98%, and will cluster closely together in the dendrogram.

Note: This functionality is not intended to infer or display familial relationships, and Illumina does not recommend using this tool for anything other than to identify samples derived from the same individuals. Replicates or matched samples will typically appear as highly correlated and are represented by closely linked branches at the far left of the dendrogram.

Figure 9. Example dendrogram based on RS‑SNP beta values, showing clustering of samples, with replicates and matched samples grouping closely together.

For any feedback or questions regarding this article (Illumina Knowledge Article #3719), contact Illumina Technical Support [e-mail].