Skip to content
For the complete documentation index, see llms.txt. This page is also available as Markdown.

Illumina Connected Multiomics Walkthrough

Illumina Connected Multiomics provides interactive visualizations and powerful statistics. This is a walkthrough of an analysis that could be done in Connected Multiomics with an example proteomic data set, produced by DRAGEN Protein Quantification. It covers the following features:

  • Creating a default analysis

  • Creating a custom analysis

  • Managing sample metadata

  • Filtering samples

  • Filtering features

  • Data Transformation

  • PCA

  • Differential expression

  • Hierarchical clustering and creating heatmaps

  • pQTL analysis

  • Correlation engine

  • GSEA

For information on the Connected Multiomics Platform, including how to log in, please reference the following documentation: https://help.multiomics.illumina.com/icm

Demo Data

Demo data that can be used to follow along with this walkthrough is found in the Connected Multiomics Demo Data repository. To add this dataset to a study, perform the following steps:

Select the data type: DRAGEN Proten Quantification

Select file format and click on Add demo data

The data used in this walkthrough can be found at /Multiomics-Demo-Data/Proteomics/NovaSeq 6k-S4 Cancer-Normal. For this study, both SampleType (CRC/Control) and TimePoint (T1...T8) are used. This data must be ingested prior to starting the analysis. Add both the ADAT (counts) and TSV (metadata) to the study.The data used in this walkthrough can be found at /Multiomics-Demo-Data/Proteomics/NovaSeq 6k-S4 Cancer-Normal. For this study, both SampleType (CRC/Control) and TimePoint (T1...T8) are used. This data must be ingested prior to starting the analysis. Add both the ADAT (counts) and TSV (metadata) to the study.

Browse to the proteomics data folder to select the adat and tsv files, click Add selected data to your study

Creating a Default Analysis

  • Select all the samples and click +Create analysis .

  • In the pop-up window, provide a name for the analysis, select Default: Proteomics as the Analysis Type, and select Run Analysis.

  • The default analysis generates a pipeline including normalization, PCA and hierarchical clustering heatmap on the imported data.

  • Double click on the Summary report to open the plots in data viewer

  • There are two pages in the viewer, the first page is Gene Expression, it contains PCA scatterplot, Scree plot, PCA component loadings table and heatmap

  • Exploring the PCA plot:

    • By default, the plot is colored by BatchID. To change this, in the left hand bar, select Configure > Style > Color by.

  • You may also want to explore other principal components. To do this, on the left side bar for the PCA plot, select configure, axes, and update the data for each axis.You may also want to explore other principal components. To do this, on the left side bar for the PCA plot, select configure, axes, and update the data for each axis.

Select the Sample Distribution page to check histogram, box-whisker plot of all the raw count distribution and scale factor distribution in each sample type.

In histogram plot, X-axis is the raw count of protein, each line is a sample

In box-whisker plot, X-axis represents samples, Y-axis represents the protein raw count

In Dotplot, X-axis represents sample type, each dot is a sample, box plot summarized the distribution of each group, Y-axis represents the HybNorm_1_ScaleFactor.

In the dotplot, choose Configure > Axes, from the Y-axis drop-down list, select different normalization scale factor (attributes in green color) to check the distribution.

Creating a Custom Analysis

Custom Analysis Example
  • Select all the samples and click +Create analysis

  • In the pop-up window, provide a name for the analysis, select Custom: Illumina Proteomics as the Analysis Type, select the sample group to be included in the analysis (All 9k SomaSeq Discovery Samples will be selected by default), and select Run Analysis.

​​

  • NOTE: Make sure there are no duplicated Sample IDs in the analysis groups.

  • A pop-up message will show up if the analysis creation is successful.

​

  • Refresh the page to get the latest status of the analysis.

  • When the Status is ‘Complete’, select the analysis tile to enter the analysis module.

​

  • There is no default initiated analysis for the custom proteomic data. To review the number of samples and features, hover over the data node.

  • Throughout the below analyses, rectangles/task nodes will produce circles/data nodes. Double-click on the task node to view details on the task as it occurred, and double-click on the data node to view the results of the analysis. Nodes will be greyed out while the analysis is still in progress.

Managing sample metadata

  • Select the Metadata tab to view and add sample metadata.

  • Select Manage under Sample attributes to reorder the metadata. Drag SampleStatus and TimePoint boxes to the front since they are the features that need to be colored for the downstream analysis. You can also add/remove/reorder other metadata or add new category to the current metadata in this page.

Filtering samples

  • Return to the analysis page, select the Quantification node, select Filtering > Filter samples from the right hand tool box.

  • Select the samples with TimePoint T1 and T2.

Filtering features

  • Select Finish and return to the analysis page. Select the TimePoint in T1,T2 node, select Filtering > Filter features from the right hand tool box.

  • Filter only include human protein by selecting the Metadata option and specify filter criteria as include Organism in Human, select Finish.

double-click the feature filter output data node to open the report, check the distribution of the data. If the min is 0, perform the Normalization task to add 1. If the min is not 0, skip the normalization step since the data is already normalized.

Normalization

  • Select the Filtered counts node, select Normalization and Scaling > Normalization from the right hand tool box.

  • Select Add and drag it to the right-hand box to avoid 0 counts. This prevents any 0 count values which could impact Limma-trend differential analysis, which assumes continuous data. Then select Finish to return to the analysis dashboard.

PCA

  • Select the Normalized counts node, select Exploratory analysis > PCA from the right hand tool box.

  • Use the default setting and select Finish.

  • double-click on the PCA node to view the PCA report.

    • The scatter plot shows the data distribution (colored by SampleStatus) among the first three PCs.

    • The scree plot (top right panel) shows the variant represented by each PC.

    • The component loading table (bottom right panel) shows the correlation between every protein/SOMAmer and each PC. The variable in this table represents the SOMAmer's SeqID.

    • For additional information on PCA, review the following documentation: https://help.partek.illumina.com/partek-flow/user-manual/task-menu/exploratory-analysis/pca

Differential expression

  • Select the Normalized counts node, select Statistics > Differential analysis from the right hand menu.

  • Select Limma-trend (default) method and select Next.

NOTE: Limma-trend is a robust model that fits the assumptions for small sample sizes of normalized protein counts. The Limma-trend model is also flexible with categorical and quantitative variables. For other datasets or experimental designs, consider other methods.

  • Select SampleStatus > TimePoint > DonorID, and then select Add factors.

  • Select SampleStatus and TimePoint, and then select Add interaction to add the factors.

  • To set up comparisons, select Next.

  • Drag CRC to the top right box and Control to the bottom right box, and then select Add comparison.

  • Select SampleStatus*TimePoint from the Factor dropdown menu. Add T1 and T2 comparison between CRC and Control. Keep Combine selected for each of these comparisons.

  • No need to perform anymore filter or normalization, so keep the Low value filter unchecked and select None for count normalization.

  • Select Finish at the bottom.

  • When the task is done, double-click on the output node to view the report. On the left hand menu

    • select FDR, select Per contrast and specify 0.05 for CRC vs Control comparison

    • select Fold change, select Per contrast and specify -2 to 2 for CRC vs Control comparison.

    • select Generate Filtered Node

    • repeat this process on CRC T1 vs Control T1 comparison and CRC T2 vs Control T2 comparison.

  • Return to the analyses dashboard and there will be 3 filtered feature list nodes added to the pipeline; right-click on the Filtered feature list node and select Rename data node to rename the node as T vs N; apply the same procedure to the other two filtered feature lists and rename them as T vs N Time 1 and T vs N Time 2 respectively.

  • To compare the filtered feature lists, select the Venn diagram on the bottom menu and tick on the filtered lists (T vs N, T vs N Time 1 and T vs N Time 2); then select Display selection on the bottom to visualize the Venn diagram.

Hierarchical clustering and creating heatmaps

  • Select the T vs N data node, select Exploratory analysis > Hierarchical clustering / heatmap from the right hand tool box.

  • Select Heatmap and select the feature order and sample order.

    • Select Cluster (default) as feature order

    • Select Assign order and select SampleStatus from the dropdown menu.

    • Select Finish at the bottom of the page.

  • double-click on the Hierarchical clustering / heatmap node to view it.

pQTL Analysis

For experiments with both protein expression and genotype information on the same set of the subjects, protein quantitative trait locus (loci) analysis can be performed in Connected Multiomics, this analysis will link a specific DNA sequence variant directly to the protein abundance. Detailed information on how to perform QTL analysis can be found in the QTL analysis chapter.

Correlation Engine

  • Select the T vs N data node, select Biological interpretation > Correlation Engine pathway from the right hand tool box.

  • Select Homo sapiens as Organism, Protein Expression as Data type, specify a name of a Correlation Engine project and study name.

  • Select Next

  • Specify Contrasts, multiple contrasts can be selected, use gene symbol as identifier. Select Finish at the bottom of the page.

  • double-click on the Correlation Engine node to view the report.

  • Select Open Data Viewer auto session to view the top 30 gene sets in Data viewer.

GSEA

  • To detect differential pathways between diseased and control samples, select the Normalized counts node (or filtered count data node) and select Biological interpretation > GSEA from the right hand tool box.

  • Select KEGG database (default) and select Next at the bottom of the page.

  • Select SampleStatus and select Next.

  • Drag CRC to the top right box and Control to the bottom right box. Keep Combine selected for this comparison. Select Add comparison.

  • No need to perform filter or normalization, select Finish on the bottom of the page.

  • double-click on the GSEA node to view the results.

  • Select the enrichment plot icon after each row index to visualize the enrichment score of the corresponding pathway.

NOTE. For clarify on the differences between Gene Set Enrichment Analysis and GSEA, please view this documentation: https://help.partek.illumina.com/partek-flow/frequently-asked-questions#what-is-the-difference-between-gsea-and-gene-set-enrichment

Last updated

Was this helpful?