Evolution of Protein Functional Annotation: Text Mining Study

Abstract

Within the Human Proteome Project initiative framework for creating functional annotations of uPE1 proteins, the neXt-CP50 Challenge was launched in 2018. In analogy with the missing-protein challenge, each command deciphers the functional features of the proteins in the chromosome-centric mode. However, the neXt-CP50 Challenge is more complicated than the missing-protein challenge: the approaches and methods for solving the problem are clear, but neither the concept of protein function nor specific experimental and/or bioinformatics protocols have been standardized to address it. We proposed using a retrospective analysis of the key HPP repository, the neXtProt database, to identify the most frequently used experimental and bioinformatic methods for analyzing protein functions, and the dynamics of accumulation of functional annotations. It has been shown that the dynamics of the increase in the number of proteins with known functions are greater than the progress made in the experimental confirmation of the existence of questionable proteins in the framework of the missing-protein challenge. At the same time, the functional annotation is based on the guilty-by-association postulate, according to which, based on large-scale experiments on API-MS and Y2H, proteins with unknown functions are most likely mapped through "handshakes" to biochemical processes.

Keywords: CHPP; Human Proteome Project; missing proteins; neXt-MP50; neXtCP-50; neXtProt; protein function; text-mining; uPE1 proteins.

Figure 1

Change in the completeness of human protein data according to neXtProt. (a) Chronology of changes in the number of protein identifications; (b) chronology of replenishment of neXtProt with information on protein functions.

Figure 2

A cloud of biological functions for proteins. The data presented refer to 1392 proteins whose functional annotation appeared in neXtProt from the beginning of 2016 to the beginning of 2021.