How reliable are experimental protein-protein interaction data?

By Einat Sprinzak, Shmuel Sattath, Hanah Margalit. Journal of Molecular Biology, 2003

440 citations recorded by the source corpus.

Subfields: Cell Biology, Molecular Biology

View Paper: https://www.semanticscholar.org/paper/45d43d4764c0211f441e0d81738711d0c3c9dcf0

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cell Biology

In Molecular Biology

Citations in the corpus

How complete are current yeast and human protein-interaction networks? (2006, background): “[24] by fitting the interaction set as a linear combination of true-positive (small scale interactions) and false-positive (random pairs) interactions. interaction assay error rates [20,22,37,41-45] and for scoring individual protein-protein interactions.”
TALE: A Tool for Approximate Large Graph Matching (2008, background): “For example, it is well known that protein interaction networks produced by high-throughput methods contain many false positives [14].”
Kernel methods for predicting protein-protein interactions (2005, background): “• Interactions that are discovered by different experimental assays were estimated to be correct 95% of the time (Sprinzak et al., 2003).”
A Comprehensive Benchmark of Kernel Methods to Extract Protein–Protein Interactions from Literature (2010, background): “These techniques are prone to produce comparably large numbers of false positives [7].”
Analyzing and modeling real-world phenomena with complex networks: a survey of applications (2007, methodology): “proposed a method to asses the quality of protein interactio n databases using the cellular localization and cellular-role properties to provide an es timative of true positives in databases [326].”
Predicting protein-protein interactions using signature products (2005, methodology): “If these data are unreliable, as is often the case with protein–protein interaction data (von Mering et al., 2002; Sprinzak et al., 2003), then the results of our algorithm will also be unreliable.”
Evaluation of clustering algorithms for protein-protein interaction networks (2006, methodology): “The error rate depends strongly on the method, high-throughput and computational methods being less reliable than traditional methods [9].”
Online Predicted Human Interaction Database (2005, positive): “For example, in mapping S. cerevisiae interactions to C. elegans , Mat-thews et al. were only able to reproduce 16 – 31% of the predicted interactions in a Y2H system.”
Whole-proteome prediction of protein function via graph-theoretic analysis of interaction maps (2005, background): “Weighting functional linkages It is well known that the reliability of different data sources vary, even if they are based on the same underlying technology ( von Mering et al., 2002; Deng et al., 2003; Sprinzak et al., 2003).”
Deciphering Protein–Protein Interactions. Part I. Experimental Techniques and Databases (2007, methodology): “Protein localization method (PLM) [98] defines true positives as interacting proteins that are localized in the same cellular compartment and/or interacting proteins that are annotated to have a common cellular role.”
Integration of biological networks and gene expression data using Cytoscape (2007, background): “For instance, it is estimated that up to 50% of unfiltered yeast two-hybrid data are spuriou”
Power-Law Distributions in Empirical Data (2007, background): “In some cases it is known that the sampling procedure used to obtain these subsets may be biased, as, for example, in the protein interactions [56], citations and authorships [9], and the Internet [2, 15].”
An empirical framework for binary interactome mapping (2008, neutral): “screening completeness of the tested spaces corresponds to: (1) (7,200 x 7200) / (22,500 x 22,500) = 10% for Space I tested in the generation of the CCSB-HI1 map, (2)”
Protein interaction networks from yeast to human. (2004, background): “However, highquality subsets could often be chosen on the basis of additional criteria [15 ,16 ,38,39], such as the degree to which mRNAs of interacting proteins are co-expressed in microarray experiments [40,41], topological properties of the resulting networks [42,43], shared pathways or subcellul”