Phyre2 is a web server to predict protein structure from sequence (www.imperial.ac.uk/phyre2) that processes ~1,000 individual sequences submitted by users every day. Since its introduction in 2011, Phyre2 has processed well over 4M jobs with ~55,000 unique users per year, each submitting ~20 sequences on average. The Phyre2 (Fig 1) web portal [1]provides both a rapid and user-friendly interface to predict protein structure using homology based template modelling and also resources for analysing the results. The papers describing Phyre2 and its predecessors (3D-PSSM and Phyre) have had over 12,000 citations in the literature.

The performance of different protein structure prediction implementations is compared in a biennial exercise, the Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction; the 14th edition, CASP14, took place in 2020 [2]. The outstanding performance of one program, AlphaFold2 [3] in CASP14 drew the attention of the world's media to this field. The results might lead the casual observer to conclude that the protein structure problem is solved, but at the moment AlphaFold2 itself is not readily accessible to the vast majority of users and the underlying methods employed have not yet been revealed in any detail.

We will show how the carefully designed interface to Phyre2 allows users to generate 3D protein structures from their sequence data in a flexible and straightforward way that makes good models readily available to the community at large.

In addition to a simple mode that allows modelling from single sequences, the Phyre2 web portal proves a range of extra functionality: i) a facility for batch submission of processing of proteomes, ii) searching model genomes for a protein structure, iii) PhyreAlarm, which automatically updates a user if a superior model can be predicted as a result of a newly-deposited structure in the protein data bank, and iv) facilities to analyse a predicted model in terms of accuracy and sequence conservation.

Phyre2 is a resource with the UK node of ELIXIR, the European–wide network of bioinformatics facilities.

Fig 1 – Main results page of the Phyre2 web server showing hits with confidence scores and origin of templates

[1] Kelley et al. (2015) Nature Protocols, 10, 845.

[2] CASP14, https://www.predictioncenter.org/casp14/index.cgi

[3] see, e.g. https://en.wikipedia.org/wiki/AlphaFold

Today the study of two-dimensional (2D) materials has become one of the key objectives of materials science. Unlike their three-dimensional counterparts, 2D materials can simultaneously demonstrate unique transport and mechanical properties due to their dimensionality and quantum size effect. Weak van der Waals interaction between layers in heterostructures of 2D materials, electron confinement inside the layers, and high surface-to-volume ratio lead to remarkable changes in electronic and optical properties of the materials, as well as in their chemical and mechanical response. Besides, a wide range of ways to tune properties using lateral and vertical heterostructures fabrication, chemical functionalization, strain, defect and substrate engineering, makes 2D materials ideal candidates for developing a new class of electronic devices. In their fabrication and application, 2D materials are usually located on top of the substrate or combined into heterostructures, which makes their structures and properties strongly depend on the nature and quality of the environment. Here, we present a novel method for studying the atomic structures of two-dimensional materials and epitaxial thin films on arbitrary substrates. The method can predict successful stages of epitaxial growth and the regions of stability of each atomic configuration with experimental parameters of interest (Figure 1). We demonstrate the performance of our methodology in the prediction of the atomic structure of MoS₂ on Al₂O₃ (0001) substrate. The method is also applied to study the CVD growth of graphene and hexagonal boron nitride on Cu (111) substrates. In both cases, stable monolayer and multilayer structures were found. The stability of all the structures in terms of partial pressures of precursors and temperature of growth is predicted within the ab initio thermodynamics approach.

Prediction of crystal structures has reached a high level of reliability, but much less is known about the mechanisms of structural transitions and pertinent barriers. The barriers related to nucleation of crystal structure inside another one are critically important for kinetics and eventually decide what structure will be created in experiment.

We show here an NPT metadynamics simulation scheme [1] employing coordination number and volume as collective variables and illustrate its application on a well-known example of reconstructive structural transformation B1/B2 in NaCl. Studying systems with size up to 64000 atoms we reach beyond the collective mechanism (Fig.1 (a)) and observe the nucleation regime (Fig.1 (b)). We reveal the structure of the critical nucleus and calculate the free-energy barrier of nucleation and also uncover details of the atomistic transition mechanism and show that it is size-dependent.

Our approach is likely to be applicable to a broader class of structural phase transitions induced by compression/decompression and could find phases unreachable by standard crystal structure prediction methods as well as reveal complex nucleation and growth effects of martensitic transitions.

Figure 1. (a) Collective mechanism - A typical frame of supercell during the course of the B1/B2 transition in NaCl at 40 GPa and 300 K in the system of size of 512 atoms. (b) Nucleation - A typical nucleus of the B2 phase in the B1 phase (with shape of ellipsoid), during the transition at 40 GPa and 300 K in the system of 64 000 atoms. Plane of view cuts the ellipsoid through its centre. Figure was produced using OVITO [2].

[1] M. Badin and R. Martoňák, arXiv:2105.02036

[2] A. Stukowski, Modelling Simul. Mater. Sci. Eng. 18, 015012 (2010).

Keywords: pressure-induced phase transitions; nucleation; martensitic transition; metadynamics

This work was supported by the Slovak Research and Development Agency under Contracts APVV-15-0496 and APVV-19-0371, by VEGA project 1/0640/20 and by Comenius University under grant for young researchers - UK/436/2021.