STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets.

Szklarczyk D, Gable AL, Lyon D, Junge A, Wyder S, Huerta-Cepas J, Simonovic M, Doncheva NT, Morris JH, Bork P, Jensen LJ, Mering CV. · 2019

Read STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. by Szklarczyk D, Gable AL, Lyon D, Junge A, Wyder S, Huerta-Cepas J, Simonovic M, Doncheva NT, Morris JH, Bork P, Jensen LJ, Mering CV. (2019) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc6323986.txt.

Introduction

Szklarczyk and co-authors announce STRING version 11, a network of protein links. Their update appeared in Nucleic Acids Research. Proteins work through webs of physical and functional ties. No single source holds all of that knowledge. STRING gathers public interaction data and scores it. It adds computer predictions to fill the gaps.

Version 11 more than doubles species coverage, to 5,090 organisms. Users can now upload whole genome-wide datasets. The site draws a network from just the uploaded genes. It also runs enrichment tests on the full input set. It uses Gene Ontology and KEGG groupings for that task. New groupings come from text mining and network clustering too. The goal is a fuller picture of how cell machinery connects.

Catalog record

CREATORSzklarczyk D, Gable AL, Lyon D, Junge A, Wyder S, Huerta-Cepas J, Simonovic M, Doncheva NT, Morris JH, Bork P, Jensen LJ, Mering CV.
YEAR2019
COLLECTIONPAPERS
FORMATTXT
FILE SIZE33 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC6323986

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

STRING v11: protein–protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets By Damian Szklarczyk, Annika L Gable, David Lyon, Alexander Junge, Stefan Wyder, Jaime Huerta-Cepas, Milan Simonovic, Nadezhda T Doncheva, John H Morris, Peer Bork, Lars J Jensen, Christian von Mering Published in: Nucleic Acids Research DOI: 10.1093/nar/gky1131 ABSTRACT Abstract Proteins and their functional interactions form the backbone of the cellular machinery. Their connectivity network needs to be considered for the full understanding of biological phenomena, but the available information on protein–protein associations is incomplete and exhibits varying levels of annotation granularity and reliability. The STRING database aims to collect, score and integrate all publicly available sources of protein–protein interaction information, and to complement these with computational predictions. Its goal is to achieve a comprehensive and objective global network, including direct (physical) as well as indirect (functional) interactions. The latest version of STRING (11.0) more than doubles the number of organisms it covers, to 5090. The most important new feature is an option to upload entire, genome-wide datasets as input, allowing users to visualize subsets as interaction networks and to perform gene-set enrichment analysis on the entire input. For the enrichment analysis, STRING implements well-known classification systems such as Gene Ontology and KEGG, but also offers additional, new classification systems based on high-throughput text-mining as well as on a hierarchical clustering of the association network itself. The STRING resource is available online at https://string-db.org/. INTRODUCTION While an impressive amount of structural and functional information on individual proteins has been amassed (1–3), our knowledge about their interactions remains more fragmented. Some interactions are quite well documented and understood, for example in the context of three-dimensional reconstructions of large cellular machineries (4–6), while others are only hinted at so far, through indirect evidence such as genetic observations or statistical predictions. Furthermore, the space of potential protein–protein interactions is much larger, and also more context-dependent, than the space of intrinsic molecular function of individual molecules. Interactions may not only be limited to certain cell types or certain physiological conditions, but their specificity and strength may vary as well, from obligatory, highly specific and stable bindings to more fleeting and relatively unspecific encounters. From a purely functional perspective, proteins can even interact specifically without touching at all, such as when a transcription factor helps to regulate the expression and production of another protein, or when two enzymes exchange a specific substrate via diffusion. Arguably, the common denominator of the various forms of protein–protein associations is information flow—biologically meaningful interfaces have evolved to allow the flow of information through the cell, and they are ultimately essential for implementing a functional system. Hence, it is desirable to collect and integrate all types of protein–protein interactions under one framework; this then provides support for data analysis pipelines in diverse areas, ranging from disease module identification (7,8) to biomarker discovery (9–11) and allows manual browsing, ad hoc discovery and annotation. Protein–protein interactions can be collected from a number of online databases (reviewed in (12,13)), as well as from individual high-throughput efforts, e.g. (14). Primary interaction databases (3,15–18) are jointly annotating experimental interaction evidence directly from the source publications, and they are coordinating their efforts through the IMEx consortium (19). They provide highly valuable added services such as curating metadata, maintaining common name spaces and devising ontologies and standards. A second source of protein–protein interaction information is provided by computational prediction efforts, some of which are hosted by dedicated databases, e.g. (20,21). Lastly, a third class of databases is dedicated to protein interactions at the widest scope, integrating both primary as well as predicted interactions, often including annotated pathway knowledge, text-mining results, inter-organism transfers or other accessory information. The STRING database (‘Search Tool for Retrieval of Interacting Genes/Proteins’) belongs to this latter class, along with GeneMania (22), FunCoup (23), I2D (24), ConsensusPathDb (25), IMP (26) and HumanNet (27)—most of which have recently been reviewed and benchmarked in (7). STRING is one of the earliest efforts (28) and strives to differentiate itself mainly through (i) high coverage, (ii) ease of use and (iii) a consistent scoring system. It currently features the largest number of organisms (5090) and proteins (24.6 million), has very broad and diverse, benchmarked data sources and provides intuitive and fast viewers for online use. It also features a number of additional data access points, such as programmatic access through an API, access through a Cytoscape app (http://apps.cytoscape.org/apps/stringapp), as well as download pages covering individual species networks and associated data. The website allows users to log on and store their searches and gene sets, and contains evidence viewers to inspect the underlying evidence of any given interaction. It also provides users with high-level information regarding their input/search data, including network enrichment statistics and functional enrichment detection, using two different conceptual frameworks for the latter (see below). Many of the features of STRING have been made available and described earlier (28–31) and the website is currently accessed by around 3500 distinct users daily; its hosting facilities have recently been replicated and placed under a commercial load balancer, to provide added stability and capacity. Users can submit multiple proteins simultaneously and visualize large networks; the Cytoscape stringApp can even handle network sizes of several thousand proteins. STRING shares its genome-, protein- and name spaces with a number of sister projects, dedicated to orthology (eggNOG (32)), small molecules (STITCH (33)), protein abundances (PaxDB (34)), tissue expression (TISSUES (35)) and viruses (Viruses.STRING (36)), respectively. Together with other online resources (including the IMEx consortium, which is one of STRING’s largest primary data sources), the STRING database has recently been awarded the status of a European Core Data Resource by ELIXIR, a pan-European bioinformatics initiative dedicated to sustainable bioinformatics infrastructure (37). As a prerequisite and consequence of this status, all interaction data and accessory information in STRING are now freely available without restrictions, under the Creative Commons Attribution (CC BY) 4.0 license. DATABASE CONTENT The basic interaction unit in STRING is the ‘functional association’, i.e. a link between two proteins that both contribute jointly to a specific biological function (38–40). For two proteins to be associated this way, they do not need to interact physically. Instead, it is sufficient if at least some part of their functional roles in the cell overlap—and this overlapping function should be specific enough to broadly qualify as a pathway or functional map (in contrast, merely sharing ‘metabolism’ as an overlapping function would be too unspecific). By this definition, even proteins that antagonize each other can be functionally associated, such as an inhibitor and an activator within the same pathway. The desired specificity cutoff for functional associations in STRING roughly corresponds to the annotation granularity of KEGG pathway maps (41), whereby maps that largely group proteins by homology (such as ‘ABC transporters’) are removed from consideration. All of the association evidence in the STRING database is categorized into one of seven independent ‘channels’: three prediction channels based on genomic context information (see below), and one channel each for (i) co-expression, (ii) text-mining, (iii) biochemical/genetic data (‘experiments’) and (iv) previously curated pathway and protein-complex knowledge (‘databases’). Users can disable all channels individually or in combinations. For each channel, separate interaction scores are available as well as

Read the complete text