SmartR is a tool for visualisation and an advanced workflows replacement, launched by a new Analysis tab.
SmartR workflows in tranSMART are highly interactive with dynamic features. Seven workflows are included in tranSMART release 16.2.
Categorical and numerical variables can be added to workflows. Numerical variables can include high-dimensional data (for example gene expression values for any specified gene).
When interacting with SmartR results the analysis is updated without the need to repeat the extraction of data from tranSMART. The data is retrieved in a "Fetch Data" step, with associated variables if needed. The data is potentially preprocessed (workflows include a preprocessing tab where options are defined).
|Correlation||Interactive correlation plot. Values can be selected or excluded and reanalysed.|
Interactive heatmap with on-the-fly clustering, categorical and numerical variables. Biomarkers can be selected from the high dimensional data. Preprocessing allows probe aggregation.
Analysis had extensive options for values to use in differential expression, column groupings, expression level and expression variability.
In visualising the results a panel at lower right allows reanalysis using alternative values, a choice of colour scheme, a choice of clustering algorithms , and zoom in/out. Mouseover displays show the data values within the heatmap. Genes in the heatmap rows can be linked to GeneCards or to EMBL-EBI Search for further information. A table below the heatmap shows gene/probe and data values in a clearly readable layout.
Analysis parameters remain available to repeat the heatmap analysis using the already fetched data.
|IpaConnector||Launch Qiagen Ingenuity Pathway Analysis (needs a demo or authorised login) to identify pathways linked to results (new 'IPA' plugin in 16.2)|
|Linegraph||Line graph workflow for time series data|
|Patientmapper||Mapping patients with categorical variables from a subset using one variable to a subset using another variable, rewriting the query in the comparison tab|
|Volcano||Volcano plot (new workflow)|
SmartR was developed by Sascha Herzinger at the University of Luxembourg and others.
Source code is in https://github.com/transmart/SmartR
SmartR requires no new configuration parameters. The plugin is included in tranSMART from release 16.2.
If you use a remote Rserve server supporting more than one tranSMART instance you can define the directory the SmartR scripts are copied to by redefining (this value shows the default):
smartR.remoteScriptDirectory = '/tmp/smart_r_scripts'
This parameter also depends on the definition of Rmodules.host (the IP address of the remote Rserve server)
In the screenshot below we have selected two subsets for study GSE8581, patients with COPD and controls. We then went to the SmartR tab, and dragged in the expression data node 'Lung' and clicked "Fetch Data" which retrieves the expression data for each subset and generates summary statistics for the cohorts and their expression data values.
We skip the PreProcess option (it has one checkbox to select "Aggregate probes") and go directly to the "Run Analysis" tab.
We change the default '100' rows to show to '20' so we have a smaller set for this example. Values under 1000 are recommended to make the output readable and to improve performance.
Leaving other options at their default values we select "Create plot".
We get a heatmap plot which we can mouse over to see information about each value, and an interactive control panel where we can add clustering (row clustering in the example) by selecting a clustering method.