Illumina Connected Multiomics
Illumina Connected Multiomics (ICM) is available for tertiary analysis of miRNA and other multiomic data.
Getting Started
Data / Task nodes and Performing tasks in Connected Multiomics
Within a study, the Analyses tab contains two elements: task nodes (rectangles) and data nodes (circles) connected by lines and arrows. Collectively, they represent a data analysis pipeline.
Clicking a data node brings up a context sensitive menu on the right. This menu changes depending on the type of data node. It will only present tasks which can be performed on that specific data type. Hover over the task to obtain additional information regarding each option.
Select the task you wish to perform from the menu. When configuring task options, additional information regarding each option is available. Click Finish to perform the task.
Depending on the task, a new data node may automatically be created and connected to the original data node. This contains the data resulting from the task. Tasks that do not produce new data types will not produce an additional data node.
To view the results of a task, click the data node and choose the Task report option on the menu.
Viewing and saving data
All data contained in data nodes can be downloaded to the local machine by selecting the node and navigating to the bottom of the toolbox then choose Download data.
The Data Viewer can be used to plot, modify, and save data. In this walkthrough the PCA data node and Hierarchical clustering / heatmap node can be automatically opened in the Data viewer by double-clicking the data node or opening the Task report from the toolbox.
To save an individual image within the Data Viewer to your machine, click Plot then Export image & select the format, size, and resolution then click Save. Use the plot-specific tools for this.
All visualizations within a sheet in the Data Viewer can be exported as one image (e.g. use one image with all plots for a poster). Use the Export drop-down at the top of the data-viewer for this and select Export image.
Import Data
microRNA data generated from secondary analysis contains sample by miRNA raw count matrix in a .txt file format. The demo data contains 6 samples from 3 different tissue types, 2 samples in each tissue type: brain, blood and liver. The libraries were sequenced on MiSeq i100 with 100M flowcell using a single read at 72bp read length with dual indexing. The samples were analyzed with the Basespace DRAGEN miRNA app and using hg38 assembly and miRBase reference v22.
First, create a study to upload data.
Click + New Study
Create a new study Add Study Name and Description
Click Create
Click + Add Data
Choose Select from ICA project
Choose Bulk > miRNA
Click the + Add Demo Data button
Select the 'Multiomics-Demo-Data' folder
Select the 'Transcriptomics' folder
Select the 'ILMN miRNA Demo data' folder
Check the 'all_samples.miRNA.UMIs.metadata.csv' and 'all_samples.miRNA.UMIs.txt' files
Click Add selected data to your study
After creating a study and adding data to the study, click + New Analysis
Give the Analysis a name, select the Analysis Type > Custom: Illumina miRNA, use all Illumina miRNA Prep samples, which is the default, and click Run Analysis
The Status will show as Complete once it is finished.
Click the Refresh button to see change the Status in real-time.
Click on the name of the analysis to open the task graph, a data node circle labled as miRNA is displayed
Hover over nodes to see details about the data node. Below, the number of samples, features, and data size is shown in the node.
Single-click on miRNA node, choose Imported count matrix report in QA/QC section to check the distribution of the data. After the task is finished, double click on output report to view it.
In the table, each row is a sample, columns are descriptive statistics metrics of the miRNA count information. It is recommended to filter out low expression miRNAs to reduce false positive in the downstream analysis, the criteria of low expression depends on the data, this table is helpful to make a decision, e.g. the median value of the matrix is 13.5, you might want to use this value as noise background value to perform low expression filter for the first iteration of downstream analysis.
Filter Low Expression miRNA
Click on the miRNA data node, choose Filtering > Filter features, select Noise reduction filter type which is default, exclude features where max is <=13.5, click Finish.
Mouse over the output Filtered counts data node to check how many features in the filtered data node. If the number is too low, you might want to redo the filter and use a more lenient filter criteria.
Annotate Features (Optional)
If you want to get genomic location information on the miRNA, you can link the data to the miRBase annotation.
Single-click the miRNA node.
Select Annotate features under the Pre-analysis tools section in the toolbox on the right.
Choose the genome and annotation files that match those used in DRAGEN then click Finish. This data is using hg38 miRBase mature microRNAs v22
In this example, we skip this step since genomic location of miRNA is not used for the downstream analysis.
Normalization and Scaling
Normalize the data to prepare for downstream analysis.
Single-click the Filtered counts node, then select the Normalization task from the Normalization and Scaling section.
Click the "Use Recommended" button or select an alternative method. We recommend the widely used Median ratio (DESeq2 only) method.
The task output Normalized counts node.
Principal Component Analysis (PCA)
Visualize sample clustering and variance.
From the Normalized Counts node, select PCA under Exploratory Analysis.
Use default settings in the dialog, click Finish.
Double click on the PCA node to view the scatterplot in Data Viewer
When color the dots with TissueType, we see separate clusters based on different tissue types, and blood samples are more different from the other two tissues.
Differential Analysis
Compare miRNA expression across experimental groups.
From the Normalized counts node, select Differential Analysis from the Statistics section.
Choose your preferred model and set up the comparison. Note that we have chosen the DESeq2 method and used the corresponding normalization prior.
Select TissueType to Add factors and click Next
Set up the three comparisons and leave other settings as default, click Finish
DESeq2 report is generated, double click on it to open the report:
Filter Feature List
Refine the list of genes/features based on criteria.
On filter panel, specify filter criteria to generate significant miRNA lists
Here is an example of generate a list of miRNA that are significantly different between liver vs blood based on FDR <=0.05, fold change up and down 2 fold.
Check FDR step up section, choose Per contrast option
Specify live vs blood value FDR <=0.05, you can uncheck other contrasts, or leave them unchanged, since default FDR <=1 is not filtering anything
Check Fold change section, choose Per contrast option
Specify exclude range from -2 to 2 for only liver vs blood
Click Generate filtered node
Option: specify similar criteria to generate significant feature list of liver vs brain and blood vs brain.
Hierarchical clustering / Heatmap
Visualize the significant miRNA list using heatmap. Select the filtered feature list based on the criteria for liver vs blood,
Select Hierarchical clustering / Heatmap from the Exploratory analysis section.
Filter only include TissueType in blood and liver, leave everything else as default, click Finish
Double-click on the output node to visualize the results in the Data viewer.
Generate Targeted mRNA List
Click on a filtered feature list, choose miRNA integration > Get targeted mRNA on the pop-up menu, use TargetScan database, click Finish.
The task generates a Targeted mRNA node, double click on it to open the report:
Context++ score: estimating the strength of repression, more negative values suggest stronger predicted repression, this is the key metric for ranking predicted targets
Context++ score percentile: percentile ranking of the score compared to all predictions, higher percentile means stronger predicted effect
Weighted Context++ score: adjusted score considering multiple sites in the same transcript
Predicted relative KD: predicted relative dissociation constant is to measure the affinity between miRNA and targeted mRNA, lower KD means stronger binding affinity, higher KD means weaker binding
Gene Set Enrichment
Identify enriched biological pathways or gene sets. Click on Targeted mRNA node
Select Gene Set Enrichment from the Biological Interpretation section.
Choose between KEGG Pathway Enrichment or Gene Set Ontology, click Finish
Double click on the output Pathway enrichment node to open the report
Filter the report based e.g. Enrichment score > 5, when the table has less than 100 rows, Data viewer plots are enabled
Click on View plots in Data Viewer to display the pathways in barchart and scatterplot
Last updated
Was this helpful?