Introduction

Comparing groups of patients with respect to biological specifics is one of the building blocks for precision medicine. Such a data-driven analysis does not have to be abstract. With VisAVis we provide a visualization tool that allows users to inspect differences between two groups of patients.

Demo can be found here: https://frankkramer-lab.github.io/VisAVis/

Technical concerns

Angular version

This project was generated with Angular CLI version 13.3.3.

Build

Build for GitHub pages according to this guide.

Data structure

To make sure, your network is compatible as input for this class comparison, check your network for the following properties

Network

The network needs to conform to the CX data model and contain the aspects nodes, edges, nodeAttributes, networkAttributes and VisAVis.

Nodes

Each node has to have an id (property @id) and a name (property n).

Edges

Each edge has to have an id (property @id), a source (property s) and a target (property ** t**).

NodeAttributes

Each node can have multiple attributes. A node attribute's relation to a node is indicated by the node attribute's property po. The name of the node attribute (property n) always starts with a patient identifier, followed by _. After that the attribute is described.

NetworkAttributes

NetworkAttributes are used to describe the patient samples. Please note: The order for each of these attributes is crucical.

• Patients: String list of patient identifiers
• PatientSubtype: String list of each patient's disease subtype
• PatientGroups: String list of each patient's group (in sum there can only be two different types of groups!)

Furthermore the property name contains the network's name that will be used as a headline for this application.

VisAVis

This aspect contains information about the network's visualization options.

• highlight: Color of the highlighted nodes' border
• properties: Network specific property definitions
• individual_properties: Properties present for every sample (i.e.g the properties are composed of ``sample-id_property, e.g. GSM150976_Score`)

The properties and individual_properties can contain the following information:

• property: Name of the visualization property (e.g. Score)
• label: Label used in the app to display the property
• type: Mapping / data type (one of continuous, discrete and boolean)
• threshold: Nodes can be selected by adjusting the threshold (only possible for continuous properties)
• mapping: Key-value-pairs defining the thresholds or values and corresponding colors.

The mapping types result in different mapping behaviors:

• Continuous: The color is mapped to a gradient with the colors corresponding to the thresholds; this property can also be used for mapping the node size
• Discrete: Only values defined in the mapping are mapped to the corresponding color
• Boolean: The matching nodes get a border in the defined color

Example

The summary network is accessible on NDEx via https://www.ndexbio.org/viewer/networks/a420aaee-4be9-11ec-b3be-0ac135e8bacf

R implementation

Supplementary information about implementing this data format as extension to the Bioconductor package RCX can be found in the R directory.