Side effect browser: the user can browse a maximum of three side effects either by name or UMLS ID. Side effects are organized by MedDRA's System and Organ Classes. In case more than one effect is selected, the resulting network only includes associated features common to all of them.

Drug browser: select a drug which causes the side effect(s). When a drug is selected, only features associated with that drug are shown.

Navigation through the biological and chemical space:

IntSide data are based on the analysis performed by Duran-Frigola & Aloy, Chem. Biol., 2013, which consists of an enrichment analysis of drug features categorized in eight levels:

Targets: Human targets with known pharmacological action from DrugBank.

Proteins: Targets and off-targets from STITCH.

Pathways: Biological human pathways from KEGG.

Processes: Biological processes from Gene Ontology.

Functions: Molecular function from Gene Ontology.

Fragments: FP4 fingerprints provided by Open Babel. 307 SMARTS patterns encode molecular structures.

Scaffolds: Scaffold network obteined through Murcko fragmentation of the full set of drug structures. Bemis & Murcko, J. Med. Chem., 1996

Structural terms: Chemical entities from ChEBI.

For each side effect-feature pair, we computed a Fisher's exact test. We defined as 'associated' those drug features with a corrected Bonferroni p-value lower than 0.05.

Side effect and drug browsers: The user can add side effects or can specify drugs to the current network.

Navigation panel: Click on a category to explore other networks related to the side effect(s).

Network view: Interactive visualization of the network. A node can be dragged and dropped.
Hold right click on a node to link to external database, select node neighbors or remove node(s).
Node size is proportional to the number of side effects in which this feature is associated. Color and width of the edges connecting an associated feature with a side effect are proportional to the Bonferroni p-value -the lower the p-value, the darker and wider the edge. Edges are labeled with an "F" when the association was also found including only "frequent" SEs in the enrichment analysis.

Layout: Modify the arrangement of the nodes in the network.

Include:

• Interacting partners: proteins that connect to other proteins.
• Disease-related genes: A side effect could be understood as a phenocopy of diseases. Drug effects might result into the same phenotype as diseases caused by genetic alterations. Genes from CTD are used in order to collect information regarding genes associated with diseases.
• Drug metabolizing enzymes: proteins which catalyzes chemical reactions involving a given drug.
• Carriers and transporters: proteins involve in the absorption, distribution and secretation of drugs. Together with enzymes, these proteins are involved in drug pharmacokinetics.
• Non-associated proteins: proteins targeted by drugs causing the effect but not associated with it according to our analysis.

Default: click to load removed nodes and edges.

Legend: Colors and symbols used in the network view.

Title: The title summarizes the contents of the displayed network.

Percentage of related drugs: This bar indicates the proportion of drugs causing the side effect and represented in the graph divided by the total number of drugs associated to the effect(s). It presents a sensitivity measure of the network displayed.

Export current view: Download network information in TSV, XML, JSON or PNG format.

- XML format description:

<graph> ROOT

<node id = "nodeID"> NODE label

<att name="class"> Node class (over-represented protein/feature, gene, side effect, drug...) </class>

<att name="name"> Node name </name>

Other attributes depending on the class of the node

</node>

<edge source = "node1 ID" target = "node2 ID"> EDGE label

<att name="interaction"> Type of interaction (ppi, over-represented feature...) </interaction>

<att name="pvalue"> Bonferroni p-value </pvalue>

</edge>

</graph>

- TSV format description: network information is presented as an interaction per row. Node attributes (ID, name, symbol, pvalue, class and size) are given in the columns.