What's new?

Over the last few months we have been working on re-coding the whole codebase of the STRING web server from the ground up. This is to make the STRING site future-proof; it will allow us to incorporate new features quicker and easier and should make the site more responsive. At this point, for you, not much should change. The site should be a bit faster and there are a couple of new features available. However, it may also temporarily make the site less stable, so forgive us if you encounter any errors. We are constantly monitoring the server's error logs and the up-time. If you have any questions about new or missing features please contact us.

Physical interactions mode Top ↑

STRING now allows you to view the physical-only interaction network. The physical network will only display edges between the proteins for which we have evidence of their binding or forming a physical complex. The network can be viewed and used the same way as the default, functional association, STRING network. It can be filtered by score or by evidence channel and is fully accessible via the API.

When you upload a set of proteins with value ("Protein with Values/Ranks" input field), STRING searches for any enrichments in your dataset using your whole dataset as a background. In addition to that STRING new correlates your input values with various whole-genome staistical measures, including:

  • average protein abundance,
  • protein length,
  • number of publications mentioning the protein in PubMed
  • predicted average protein disorder,
  • average GC content of the encoding transcript.

All the plots are automatically generated, which lets you uncover potential trends in your data at a glance that may have an explanatory role in your enrichment analysis.

STRING cluster enrichment Top ↑

Each time you query STRING with a set of proteins we automatically, in the background, check for any possible enrichment in your data. This includes KEGG pathways, UniProt keywords, Gene Ontology terms etc. and now also STRING local neighborhood clusters. These clusters are derived from hierarchically clustering the full STRING network. Such clustering generates sets of functionally associated proteins on multiple levels of hierarchy ranging from small groups of 5 proteins to large assembles of 200 proteins. Each of these overlapping clusters is then named based on the annotations of its proteins. One advantage that STRING clusters have traditional enrichment categories is that they cover less studied, potentially not well annotated, parts of the networks.

Query names in the network Top ↑

A feature that has been requested for a long time already is the ability for STRING to display the user provided gene names in the network view, instead of the official names (or locus identifiers if an official name is not available) that STRING parsers from the source databases. Now you can choose for STRING to display your query names, instead of the default names in the "settings" tab. For proteins that are not part of your query (like the added interaction partners), STRING will default to the official gene names.