CanIsoNet: Disease-specific Isoform Interaction Network

CanIsoNet serves as a gateway to understand the functional role of disease-specific isoform switching events. We provide two main functionalities for the users:
1. Browsing various statistics on disease-specific isoform switching events
2. Exploring and visualising isoform-specific network disruptions in the STRING interaction network

Starting Points at Home Page:

CanIsoNet offers two starting points to the user:

1) Browse Disease Type: The user can find 25 disease types to browse. In the table, the first column represents a PCAWG-Code used in the Pan-Cancer Analysis Whole Genome Project for 24 cancer types and one BioProject Accession Number for OIH. The second column corresponds to the name of the disease type. The third column indicates the data source (project).

2) Browse a Genes & Isoforms: The user can search for a specific gene name or ENST id, or corresponding transcript name. Isoform search leads the user to the Transcript Page while gene search leads the user to the Gene page.

Example Use Case: Browse Disease Type Click on any of the column to be directed into the disease page (Figure A).

Disease Page

Browsing by disease type directs users to a disease page. This page presents three main findings:
a. In the Figure B-left, you will see frequencies of the top ten most common cMDTs, which represents the proportion of the transcripts occurence across cancer samples. Some cMDTs are found in every sample, which makes them potential biomarkers for that cancer type.

b. In the Figure B-right: you will find the distribution of cMDT counts across samples of the cancer type selected which is also presented as a table on the bottom table of the page.

c. In the table below the figures, the user can find all different isoforms found as cMDTs in the cancer type of interest. The user can search for a specific isoform or sample.

Transcript Page

The transcript page contains the following information:
a. STRING-based transcript interaction network (Figure D):
- Up to 20 interactions of the protein of interest are shown in the network.
- The missing interactions upon isoform switching are shown as a thick black lines. The remaining interactions are seen as a thin grey lines.
- Proteins found in the cancer gene census are highlighted in turquoise colours.

b.Statistics on the cMDT (Figure E-F): Number of missed and remaining interactions, network density score ( which estimates the density of interactions of a protein and its local neighborhood in the STRING interaction network), and cancer types where the transcript is found as cMDT.

c.Figure G-top shows the table that includes the sample-based information for the interactions lost upon isoform switching: which interactions are lost in which sample.

d.Figure G-bottom shows the table that includes the cancer census genes, included if an interaction with the protein of interest is lost upon isoform switching.

Sample Page

In sample based page, the user can see the MDT-cMDT switch in the corresponding sample. Here, we also show the TPM counts for the MDTs and cMDTs in cancer sample along with the average TPM counts of MDTs and cMDTs in normal samples (Figure H). In addition, domain structure of MDT and cMDT are provided.

Under normal conditions, cells of almost all tissue types express the same predominant canonical transcript isoform at each gene locus (MDT). In cancer, however, splicing regulation is often disturbed, leading to a switch from the most dominant transcripts (MDT) to a cancer-specific MDT (cMDT) (see Figure below).
To understand to which extent of these switches occur, we have recenty analyzed isoform-specific protein–protein interaction disruptions in 1,209 cancer samples covering 27 different cancer types from the Pan-Cancer Analysis of Whole Genomes (PCAWG) project of the International Cancer Genomics Consortium (ICGC).
Our study revealed large variations in the number of cMDT. Some cMDT were found in 100% of all samples in a cancer type, making them ideal candidates for diagnostic biomarkers.

Figure. Representation of an Isoform Switching Event: Most Dominant Transcript (MDT) is represented as red transcript in normal samples while it is switched to dark gray one in cancer samples.

The methodology behind CanIsoNet is as following:

Figure. Overview of methodology to assess the impact of cancer-specific Most Dominant Transcripts (cMDT) using an isoform-specific interaction network. The top shows the steps and filters for cMDT detection. The bottom describes the methods and databases of the isoform-specific interaction network CanIsoNet. The central section depicts the combination of cMDT information with data from CanIsoNet to assess the functional impact of alternatively spliced isoforms.

The sample size used to assess cMDTs in PCAWG and GTEx as following:

CanIsoNet currently only reports the genes and corresponding isoforms that have been detected as disease-specific isoform in our analysis. If the protein/isoform of interest is not found in CanIsoNet, that means we do not see it as disease-specific cMDT in 24 different cancer types analysis and 1 OIH data analysis.
We would like to hear them all!
Please contact us via
We will get back to you as soon as possible :)
In CanIsoNet, we only show protein interactions whose domain-domain interaction information is available. If there is no domain information available, they are discarded in analysis.
Moreover, tables in the transcript page show missing interactions if both interaction partner is expressed in a specific disease type. If proteins have not found in cancer samples (not expressed), we can not find them as interaction lost. Thus, they are not seen in the table.

If you use CanIsoNet, please cite the following articles :
Kahraman et al., Pathogenic impact of transcript isoform switching in 1,209 cancer samples covering 27 cancer types using an isoform-specific interaction network, Scientific Reports, 2020.
Karakulak et al., CanIsoNet: A Database to Study the Functional Impact of Isoform Switching Events in Cancer, BioRxiv, 2021.

If you have any inquires about CanIsoNet:

Please contact via

Follow us:

Group Page