Sense and simplicity in HADDOCK scoring: Lessons from CASP-CAPRI round 1.

Our information-driven docking approach HADDOCK is a consistent top predictor and scorer since the start of its participation in the CAPRI community-wide experiment. This sustained performance is due, in part, to its ability to integrate experimental data and/or bioinformatics information into the modelling process, and also to the overall robustness of the scoring function used to assess and rank the predictions. In the CASP-CAPRI Round 1 scoring experiment we successfully selected acceptable/medium quality models for 18/14 of the 25 targets - a top-ranking performance among all scorers. Considering that for only 20 targets acceptable models were generated by the community, our effective success rate reaches as high as 90% (18/20). This was achieved using the standard HADDOCK scoring function, which, thirteen years after its original publication, still consists of a simple linear combination of intermolecular van der Waals and Coulomb electrostatics energies and an empirically derived desolvation energy term. Despite its simplicity, this scoring function makes sense from a physico-chemical perspective, encoding key aspects of biomolecular recognition. In addition to its success in the scoring experiment, the HADDOCK server takes the first place in the server prediction category, with 16 successful predictions. Much like our scoring protocol, because of the limited time per target, the predictions relied mainly on either an ab initio center-of-mass and symmetry restrained protocol, or on a template-based approach whenever applicable. These results underline the success of our simple but sensible prediction and scoring scheme. Proteins 2017; 85:417-423. © 2016 Wiley Periodicals, Inc.

The DisVis and PowerFit Web Servers: Explorative and Integrative Modeling of Biomolecular Complexes.

Structure determination of complex molecular machines requires a combination of an increasing number of experimental methods with highly specialized software geared toward each data source to properly handle the gathered data. Recently, we introduced the two software packages PowerFit and DisVis. These combine high-resolution structures of atomic subunits with density maps from cryo-electron microscopy or distance restraints, typically acquired by chemical cross-linking coupled with mass spectrometry, respectively. Here, we report on recent advances in both GPGPU-accelerated software packages: PowerFit is a tool for rigid body fitting of atomic structures in cryo-electron density maps and has been updated to also output reliability indicators for the success of fitting, through the use of the Fisher z-transformation and associated confidence intervals; DisVis aims at quantifying the information content of distance restraints and identifying false-positive restraints. We extended its analysis capabilities to include an analysis of putative interface residues and to output an average shape representing the putative location of the ligand. To facilitate their use by a broad community, they have been implemented as web portals harvesting both local CPU resources and GPGPU-accelerated EGI grid resources. They offer user-friendly interfaces, while minimizing computational requirements, and provide a first interactive view of the results. The portals can be accessed freely after registration via http://milou.science.uu.nl/services/DISVIS and http://milou.science.uu.nl/services/POWERFIT.

Defining the limits and reliability of rigid-body fitting in cryo-EM maps using multi-scale image pyramids.

Cryo-electron microscopy provides fascinating structural insight into large macromolecular machines at increasing detail. Despite significant advances in the field, the resolution of the resulting three-dimensional images is still typically insufficient for de novo model building. To bridge the resolution gap and give an atomic interpretation to the data, high-resolution models are typically placed into the density as rigid bodies. Unfortunately, this is often done manually using graphics software, a subjective method that can lead to over-interpretation of the data. A more objective approach is to perform an exhaustive cross-correlation-based search to fit subunits into the density. Here we show, using five experimental ribosome maps ranging in resolution from 5.5 to 6.9Å, that cross-correlation-based fitting is capable of successfully fitting subunits correctly in the density for over 90% of the cases. Importantly, we provide indicators for the reliability and ambiguity of a fit, using the Fisher z-transformation and its associated confidence intervals, giving a formal approach to identify over-interpreted regions in the density. In addition, we quantify the resolution requirement for a successful fit as a function of the subunit size. For larger subunits the resolution of the data can be down-filtered to 20Å while still retaining an unambiguous fit. We leverage this information through the use of multi-scale image pyramids to accelerate the search up to 30-fold on CPUs and 40-fold on GPUs at a negligible loss in success rate. We implemented this approach in our rigid-body fitting software PowerFit, which can be freely downloaded from https://github.com/haddocking/powerfit.

The HADDOCK2.2 Web Server: User-Friendly Integrative Modeling of Biomolecular Complexes.

The prediction of the quaternary structure of biomolecular macromolecules is of paramount importance for fundamental understanding of cellular processes and drug design. In the era of integrative structural biology, one way of increasing the accuracy of modeling methods used to predict the structure of biomolecular complexes is to include as much experimental or predictive information as possible in the process. This has been at the core of our information-driven docking approach HADDOCK. We present here the updated version 2.2 of the HADDOCK portal, which offers new features such as support for mixed molecule types, additional experimental restraints and improved protocols, all of this in a user-friendly interface. With well over 6000 registered users and 108,000 jobs served, an increasing fraction of which on grid resources, we hope that this timely upgrade will help the community to solve important biological questions and further advance the field. The HADDOCK2.2 Web server is freely accessible to non-profit users at http://haddock.science.uu.nl/services/HADDOCK2.2.

DisVis: quantifying and visualizing accessible interaction space of distance-restrained biomolecular complexes.

UNLABELLED: We present DisVis, a Python package and command line tool to calculate the reduced accessible interaction space of distance-restrained binary protein complexes, allowing for direct visualization and quantification of the information content of the distance restraints. The approach is general and can also be used as a knowledge-based distance energy term in FFT-based docking directly during the sampling stage. AVAILABILITY AND IMPLEMENTATION: The source code with documentation is freely available from https://github.com/haddocking/disvis. CONTACT: a.m.j.j.bonvin@uu.nl SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Defining the limits of homology modeling in information-driven protein docking.

Information-driven docking is currently one of the most successful approaches to obtain structural models of protein interactions as demonstrated in the latest round of CAPRI. While various experimental and computational techniques can be used to retrieve information about the binding mode, the availability of three-dimensional structures of the interacting partners remains a limiting factor. Fortunately, the wealth of structural information gathered by large-scale initiatives allows for homology-based modeling of a significant fraction of the protein universe. Defining the limits of information-driven docking based on such homology models is therefore highly relevant. Here we show, using previous CAPRI targets, that out of a variety of measures, the global sequence identity between template and target is a simple but reliable predictor of the achievable quality of the docking models. This indicates that a well-defined overall fold is critical for the interaction. Furthermore, the quality of the data at our disposal to characterize the interaction plays a determinant role in the success of the docking. Given reliable interface information we can obtain acceptable predictions even at low global sequence identity. These results, which define the boundaries between trustworthy and unreliable predictions, should guide both experts and nonexperts in defining the limits of what is achievable by docking. This is highly relevant considering that the fraction of the interactome amenable for docking is only bound to grow as the number of experimentally solved structures increases.