Negatome 2.0: a database of non-interacting proteins derived by literature mining, manual annotation and protein structure analysis

Philipp Blohm,Florian Goebels,Benedikt Wachinger,Goar Frishman,Andreas Ruepp,Pawel Smialowski,Dmitrij Frishman

doi:10.1093/nar/gkt1079

Philipp Blohm, Florian Goebels + Show 5 more

Open Access

https://doi.org/10.1093/nar/gkt1079

Copy DOI

Abstract

Knowledge about non-interacting proteins (NIPs) is important for training the algorithms to predict protein–protein interactions (PPIs) and for assessing the false positive rates of PPI detection efforts. We present the second version of Negatome, a database of proteins and protein domains that are unlikely to engage in physical interactions (available online at http://mips.helmholtz-muenchen.de/proj/ppi/negatome). Negatome is derived by manual curation of literature and by analyzing three-dimensional structures of protein complexes. The main methodological innovation in Negatome 2.0 is the utilization of an advanced text mining procedure to guide the manual annotation process. Potential non-interactions were identified by a modified version of Excerbt, a text mining tool based on semantic sentence analysis. Manual verification shows that nearly a half of the text mining results with the highest confidence values correspond to NIP pairs. Compared to the first version the contents of the database have grown by over 300%.

Highlights

Extensive protein interaction maps have been derived for a number of model organisms by modern high-throughput techniques such as yeast two-hybrid assay
We present the second version of Negatome, a database of proteins and protein domains that are unlikely to engage in physical interactions
Manual verification shows that nearly a half of the text mining results with the highest confidence values correspond to non-interacting proteins (NIPs) pairs

Summary

INTRODUCTION

Extensive protein interaction maps have been derived for a number of model organisms by modern high-throughput techniques such as yeast two-hybrid assay. The overlap between different experimental datasets is quite poor, indicating that experimental methods possess characteristics biases and capture molecular interactions only partially This means that just because two proteins have not yet been reported as interacting does not mean that they do not interact in the cell. Nucleic Acids Research, 2014, Vol 42, Database issue D397 non-interacting pairs were derived from these two datasets by excluding interactions detected by high-throughput approaches (1162 literature-derived and 745 structurederived negative interactions, respectively). It spite of the Negatome’s obvious bias toward well-studied cases described in literature and documented by 3D structure analysis [7], it has become a useful tool in PPI analysis and prediction. Negatome 2.0 comprises all NIPs from Negatome 1.0 and the additional NIPs that were derived as described in the following

DETECTING NEGATIVE PROTEIN INTERACTIONS BY TEXT MINING

The PDB database

Number of pairs

Findings

MANUAL CURATION AND VERIFICATION OF TEXT MINING RESULTS