PPI Database: Why STRING Is Significant?

PPI databases unambiguously expands our opportunity to figure out a fruit for a PPI drug target. Currently, more than twenty databases can be utilized for the sake of PPI research and predictions. One of the topmost databases on PPI is, generally speaking, STRING.

PPI scientists usually takes STRING as one of the reliable, versatile and useful PPI database, but have you ever thought why STRING? Emerging PPI databases may also be worth looking at.

This paper revealed the reason and gives you better understanding of the user-chosen 16 databases.²⁾

PPI data enclosed in the databases are divided into two classes: experimentally verified and predicted. Owing to the development of computational technology and the improvement of calculation speed, prediction of interaction is currently possible choice of novel PPI search and even drug target selection.

But it is obvious that experimentally validated PPIs are more fascinating candidate to throw tremendous resources for drug development.

HT (high-throughput) approaches have been developed, and yeast two-hybrid (Y2H) system³⁾ or affinity purification followed by mass spectrometry (AP-MS)⁴⁾ are mainly used so far. These two methodologies have greatly contributed to the expansion of reliable PPI datasets.⁵⁾

STRING is significant, at least for pharmaceutical companies, because of the number of validated PPI data. STRING does include predicted PPI but around 80% of the data are experimentally validated, although each experiment has its own limitation.
hPRINT would be more interesting database if you want to know the human PPI prediction data since the number of predicted interactions of hPRINT surpasses that of STRING.

But STRING is not THE database for PPI. Systematic comparison of the 16 databases revealed that BioGRID is the experimentally verified, primary-source PPI database that has the best coverage of literatures curated PPI.

Among the primary PPI databases, BioGRID contains more proteins and exclusive interactions than others as well. In a sense, BioGRID would be the golden sea of PPI drug target that is not prevalently known even by STRING.

STRING is expanding and currently the numbers of proteins on the database are 67.6 million and those of the interaction are over 20 billion. STRING is optimum for the thorough search of experimental and predicted PPI. However, databases like hPRINT and BioGRID works better in some cases.

As a leading company of peptidomimetics-based PPI drug discovery, we are pretty interested in PPI databases and utilizing them for our research activities.

We would say the know-how of the latest PPI databases would contribute your problem-solving. Please feel free to ask us for discussion and we’ll definitely try to figure out the way when you are targeting PPI for drug discovery and development.

1) https://string-db.org/
2) https://doi.org/10.1016/j.jbi.2020.103380
3) https://doi.org/10.1038/340245a0
4) https://doi.org/10.1038/13732
5) https://doi.org/10.1186/1479-7364-3-3-291