DEPICTER: Intrinsic Disorder and Disorder Function Prediction Server

Highlights

• New DisorderEd PredictIon CenTER (DEPICTER) server was developed.

• DEPICTER provides accurate predictions of disorder and disorder functions.

• DEPICTER covers protein and nucleic acids binding, linker, and moonlighting functions.

• DEPICTER makes predictions in under 1 min for an average-size protein sequence.

Abstract

Computational predictions of the intrinsic disorder and its functions are instrumental to facilitate annotation for the millions of unannotated proteins. However, access to these predictors is fragmented and requires substantial effort to find them and to collect and combine their results. The DEPICTER (DisorderEd PredictIon CenTER) server provides first-of-its-kind centralized access to 10 popular disorder and disorder function predictions that cover protein and nucleic acids binding, linkers, and moonlighting regions. It automates the prediction process, runs user-selected methods on the server side, visualizes the results, and outputs all predictions in a consistent and easy-to-parse format. DEPICTER also includes two accurate consensus predictors of disorder and disordered protein binding. Empirical tests on an independent (low similarity) benchmark dataset reveal that the computational tools included in DEPICTER generate accurate predictions that are significantly better than the results secured using sequence alignment. The DEPICTER server is freely available at http://biomine.cs.vcu.edu/servers/DEPICTER/.

Introduction

Intrinsic disordered proteins (IDPs) and intrinsically disordered protein regions (IDRs) lack stable tertiary structure and form dynamic conformational ensembles under physiological conditions [[1], [2], [3]]. Recent computational studies estimate that they are highly abundant in nature, with up to 17% of eukaryotic proteins (depending on an organism) that are entirely disordered [4] and between 30 and 50% that have at least one long IDR (≥30 consecutive residues) [5,6]. IDPs and IDRs are instrumental for a wide range of cellular functions that include signaling and molecular recognition [7,8], translation [[9], [10], [11]], regulation of transcription [12], cell death processes [[13], [14], [15]], innate immune response [16], viral life cycle [[17], [18], [19]], and many others. They are implicated in the dark proteome [20,21], often found to contribute to human diseases [22,23], and are being considered as attractive new class of targets for drug discovery [24,25]. However, experimental annotations of IDRs and their functions are limited to only about 1600 proteins that are stored in the DisProt database [26,27]. This gave rise to the development of a large collection of over 70 computational methods that predict IDRs and IDPs from the protein sequences [3,[28], [29], [30], [31], [32], [33]]. Recent empirical studies have shown that some of these methods provide highly accurate predictions [[34], [35], [36], [37]]. Moreover, two dozen computational tools that predict several functions of IDRs were published and released over the last decade [32,38,39]. These methods address sequence-based prediction of molecular partners that interact with IDRs, including proteins, DNA and RNAs, and a selected set of cellular functions, such as flexible linkers and moonlighting regions. These computational predictors can be used to accurately and in cost- and runtime-efficient way predict and functionally annotate IDPs and IDRs for the millions of proteins that lack these annotations.

The predictions produced by these methods can be collected either by the means of their webservers/implementations provided and supported by the authors and by using popular and large databases of precomputed predictions: D²P² [40] and MobiDB [41]. Both databases offer access to results generated by a large pool of disorder predictors for millions of sequences proteins. More specifically, D²P² covers 9 disorder predictions for about 10.5 million proteins, while MobiDB provides 10 disorder predictions for the contents of the October 2017 version of the UniProt repository [42], which includes around 90 million proteins. While these two resources provide unrivaled coverage of the disorder predictions, they offer a rather limited selection of the disorder function predictions that includes only the protein-binding regions predicted by the ANCHOR method [43]. Moreover, they are constrained to the set of proteins that they currently include. This means that the users must use the webservers or implementations of the available predictors for the huge number of proteins that are not currently included in these databases. More specifically, the current version of UniProt includes over 171 million proteins compared to 90 million in MobiDB. Collection of these predictions is rather difficult since it demands finding their websites/implementations, running the predictions using multiple different interfaces that require reformatting the input protein data, and assembling the results that are provided in a wide range of formats. The aforementioned issues can be alleviated by the development of a predictive resource that provides integrated access to a comprehensive set of disorder and disorder function predictors. While there are no such resources for the disorder prediction, they are already available for the prediction of various aspects of protein structure, including PSIPRED workbench [44], SCRATCH [45], PredictProtein [46], and MULTICOM [47].

To this end, we provide first-of-its-kind webserver for the sequence-based prediction of intrinsic disorder and disorder functions. The DEPICTER (DisorderEd PredictIon CenTER) server integrates predictions of intrinsic disorder and several disorder functions using several popular and runtime-efficient methods. The server automates the entire process of prediction across all these tools, visualizes the results, and outputs an easy-to-parse text file that provides all predictions in a consistent format. DEPICTER is freely available at http://biomine.cs.vcu.edu/servers/DEPICTER/.