The purpose of this app is to construct a network by co-occurence analysis. The theory behind is there is some connection if two items share the same profile, known as co-occurence. This app helps you analyze correlation and generate cooccurrence visualizations between proteins, functions, bacteria or other features in your dataset. Graphic visualization of potential relationships can be presented using network, circus, sankey plots, etc.
This shiny app uses visNetwork package for displaying network, for detailed information please refer to visNetwork. More R packages include: shiny, shinydashboard, htmlwidgets, DT, ggplot2, plotly, corrplot, colourpicker, circlize.
Step 1. Prepare your data
In order to make the app work for you properly, it is important to prepare your data input correctly.
Two files are required for the current version. The first one is the expression matrix, with rows as features, and columns as samples. The second one is the grouping information for each feature. This information would be helpful for coloring different groups in network construction.
An example is given below, the data matrix includes protein and COG features, and their intensities in each sample. The meta table marks which features are proteins and which are COGs.
Step 2. Upload data and go to analysis
Go to https://shiny.imetalab.ca/matrix2network/ and upload your data, then click “Go to Network Analysis” button. This will lead you to the network construction settings page.
You could select correlation method between Spearman and Pearson’s correlations, as well as analysis thresholds. After settings, click “Network Construction and Display” button, and a network will show up in the box below it.
You can play around with the plotting parameters and click
“Network Construction and Display” button again to update the figure.
Click “Save network” button at the bottom-right of the figure to save as PNG.
Step 3. View more results
After the network is successfully constructed, a “View more result” button appears in the box.
Click the button and you can obtain Circus, Sankey, Correlation plots, and download the results.