Towards the automatic crystal structure solution of nucleic acids: automated model building using the new CAB program

CAB, a recently described automated model-building (AMB) program, has been modified to work effectively with nucleic acids. To this end, several new algorithms have been introduced and the libraries have been updated. To reduce the input average phase error, ligand heavy atoms are now located before starting the CAB interpretation of the electron-density maps. Furthermore, alternative approaches are used depending on whether the ligands belong to the target or to the model chain used in the molecular-replacement step. Robust criteria are then applied to decide whether the AMB model is acceptable or whether it must be modified to fit prior information on the target structure. In the latter case, the model chains are rearranged to fit prior information on the target chains. Here, the performance of the new AMB program CAB applied to various nucleic acid structures is discussed. Other well documented programs such as Nautilus, ARP/wARP and phenix.autobuild were also applied and the experimental results are described.

Protein identification from electron cryomicroscopy maps by automated model building and side-chain matching

Using single-particle electron cryo-microscopy (cryo-EM), it is possible to obtain multiple reconstructions showing the 3D structures of proteins imaged as a mixture. Here, it is shown that automatic map interpretation based on such reconstructions can be used to create atomic models of proteins as well as to match the proteins to the correct sequences and thereby to identify them. This procedure was tested using two proteins previously identified from a mixture at resolutions of 3.2 Å, as well as using 91 deposited maps with resolutions between 2 and 4.5 Å. The approach is found to be highly effective for maps obtained at resolutions of 3.5 Å and better, and to have some utility at resolutions as low as 4 Å.

Predicting protein model correctness in Coot using machine learning

Manually identifying and correcting errors in protein models can be a slow process, but improvements in validation tools and automated model-building software can contribute to reducing this burden. This article presents a new correctness score that is produced by combining multiple sources of information using a neural network. The residues in 639 automatically built models were marked as correct or incorrect by comparing them with the coordinates deposited in the PDB. A number of features were also calculated for each residue using Coot, including map-to-model correlation, density values, B factors, clashes, Ramachandran scores, rotamer scores and resolution. Two neural networks were created using these features as inputs: one to predict the correctness of main-chain atoms and the other for side chains. The 639 structures were split into 511 that were used to train the neural networks and 128 that were used to test performance. The predicted correctness scores could correctly categorize 92.3% of the main-chain atoms and 87.6% of the side chains. A Coot ML Correctness script was written to display the scores in a graphical user interface as well as for the automatic pruning of chains, residues and side chains with low scores. The automatic pruning function was added to the CCP4i2 Buccaneer automated model-building pipeline, leading to significant improvements, especially for high-resolution structures.

Current approaches for automated model building into cryo-EM maps using Buccaneer with CCP-EM

This work focuses on the use of the existing protein-model-building software Buccaneer to provide structural interpretation of electron cryo-microscopy (cryo-EM) maps. Originally developed for application to X-ray crystallography, the necessary steps to optimise the usage of Buccaneer with cryo-EM maps are shown. This approach has been applied to the data sets of 208 cryo-EM maps with resolutions of better than 4 Å. The results obtained also show an evident improvement in the sequencing step when the initial reference map and model used for crystallographic cases are replaced by a cryo-EM reference. All other necessary changes to settings in Buccaneer are implemented in the model-building pipeline from within the CCP-EM interface (as of version 1.4.0).

Cyclic Automated Model Building (CAB) Applied to Nucleic Acids

Obtaining high-quality models for nucleic acid structures by automated model building programs (AMB) is still a challenge. The main reasons are the rather low resolution of the diffraction data and the large number of rotatable bonds in the main chains. The application of the most popular and documented AMB programs (e.g., PHENIX.AUTOBUILD, NAUTILUS and ARP/wARP) may provide a good assessment of the state of the art. Quite recently, a cyclic automated model building (CAB) package was described; it is a new AMB approach that makes the use of BUCCANEER for protein model building cyclic without modifying its basic algorithms. The applications showed that CAB improves the efficiency of BUCCANEER. The success suggested an extension of CAB to nucleic acids—in particular, to check if cyclically including NAUTILUS in CAB may improve its effectiveness. To accomplish this task, CAB algorithms designed for protein model building were modified to adapt them to the nucleic acid crystallochemistry. CAB was tested using 29 nucleic acids (DNA and RNA fragments). The phase estimates obtained via molecular replacement (MR) techniques were automatically submitted to phase refinement and then used as input for CAB. The experimental results from CAB were compared with those obtained by NAUTILUS, ARP/wARP and PHENIX.AUTOBUILD.

Sequence assignment for low-resolution modelling of protein crystal structures

The performance of automated model building in crystal structure determination usually decreases with the resolution of the experimental data, and may result in fragmented models and incorrect side-chain assignment. Presented here are new methods for machine-learning-based docking of main-chain fragments to the sequence and for their sequence-independent connection using a dedicated library of protein fragments. The combined use of these new methods noticeably increases sequence coverage and reduces fragmentation of the protein models automatically built with ARP/wARP.

Phasing with calcium at home

With better tools for data processing and with synchrotron beamlines that are capable of collecting data at longer wavelengths, sulfur-based native single-wavelength anomalous dispersion (SAD) phasing has become the `first-choice' method for de novo protein structure determination. However, for many proteins native SAD phasing can be simplified by taking advantage of their interactions with natural metal cofactors that are stronger anomalous scatterers than sulfur. This is demonstrated here for four unique domains of a 1.5 MDa calcium-dependent adhesion protein using the anomalous diffraction of the chelated calcium ions. In all cases, low anomalous multiplicity X-ray data were collected on a home-source diffractometer equipped with a chromium rotating anode (λ = 2.2909 Å). In all but one case, calcium SAD phasing alone was sufficient to allow automated model building and refinement of the protein model after the calcium substructure had been determined. Given that Ca atoms will be present in a significant percentage of proteins that remain uncharacterized, many aspects of the data-collection and processing methods described here could be broadly applied for routine de novo structure elucidation.