Themed collection Computational protein design and structure prediction: Celebrating the 2024 Nobel Prize in Chemistry

Proposed de novo peptide design strategy against amyloidogenic targets. After initial computational preparation of the binder and target, the computational and experimental validation are incorporated in iterative machine learning powered cycles to generate better and improved peptide-based targets.

Enzymes exist as a dynamic ensemble of conformations, each potentially playing a key role in substrate binding, the chemical transformation, or product release. We discuss recent advances in the evaluation of the enzyme conformational dynamics and its evolution towards new functions or substrate preferences.

Structures of intact polyketide synthase modules reveal conformational rearrangements and suggest asynchronous use of reaction chambers.

Efficiently harnessing big data by combining molecular modelling and machine learning accelerates rational enzyme design for its applications in fine chemical synthesis and waste valorization, to address global environmental issues and sustainable development.

Shotgun metagenomic approaches to uncover new enzymes are underdeveloped relative to PCR- or activity-based functional metagenomics. Here we review computational and experimental strategies to discover biosynthetic enzymes from metagenomes.

This review illustrates the current state in designing coiled-coil-based proteins with an emphasis on coiled coil protein origami structures and their potential.

Using structure and sequence based analysis we can engineer proteins to increase their thermal stability.

Enzymes can be optimized to accelerate chemical transformations via a range of methods. In this review, we showcase how protein engineering and computational design techniques can be interfaced to develop highly efficient and selective biocatalysts.

Rosetta design software was employed to remodel the substrate specificity of Drosophila melanogaster 2′-deoxyribonucleoside kinase for efficient phosphorylation of the nucleoside analog prodrug 3′-deoxythymidine.

De novo designed peptides bind specific conformers of α-synuclein fibrils.

Hydrogen atom transfer (HAT) reactions, as they occur in many biological systems, are here predicted by machine learning.

Pairing experiments with simulations, we predict spectroscopic fingerprints, enhancing understanding of disordered peptides' conformational ensembles. This helps rationalize elusive structure-spectra relationships for these peptides and proteins.

Structure-based three-parameter model that integrates local binding site information, coevolutionary information, and information on dynamic allostery to identify potentially hidden allosteric sites in ensembles of protein structures.

We develop a computational method integrating a genetic algorithm with a residue-level coarse-grained model of intrinsically disordered proteins in order to uncover the molecular origins of multiphase condensates and enable their controlled design.

A novel CDK20 small molecule inhibitor discovered by artificial intelligence based on an AlphaFold-predicted structure demonstrates the first application of AlphaFold in hit identification for efficient drug discovery.

Comparing the different methods for determining oligomerization composition of a protein in solution at different concentrations. The ruler of μg ml⁻¹ represents protein concentrations applicable for the different methods.

Rules for designing 4-helix bundles are defined, tested, and used to generate de novo peptide assemblies and a single-chain protein.

The dominant binding mode of the QUB-00006-Int-07 main protease inhibitor during absolute binding free energy simulations.

We generate 3D molecules conditioned on receptor binding sites by training a deep generative model on protein–ligand complexes. Our model uses the conditional receptor information to make chemically relevant changes to the generated molecules.

We used computational design to increase quantum yield in a fluorescent protein by optimizing chromophore packing to reduce non-radiative decay, resulting in an >10-fold increase in quantum yield that was further improved by directed evolution.

Molecular dynamics based absolute protein–ligand binding free energies can be calculated accurately and at large scale to facilitate drug discovery.

Antibody therapeutics and vaccines are among our last resort to end the raging COVID-19 pandemic.

A hierarchical computational strategy for IDP drug virtual screening (IDPDVS) was proposed and successfully applied to identify compounds that bind p53 TAD1 and restore wild-type p53 function in cancer cells.

Using reversed allosteric communication, we performed MD simulations, MSMs, and mutagenesis experiments, to discover allosteric sites. It reproduced the known allosteric site for MDL-801 on Sirt6 and uncovered a novel cryptic allosteric Pocket X.

De novo enzymes capable of efficiently catalysis of a non-natural reaction are obtained through minimalist design plus computationally-focused variant library screening.

The DEEPScreen system is composed of 704 target protein specific prediction models, each independently trained using experimental bioactivity measurements against many drug candidate small molecules, and optimized according to the binding properties of the target proteins.

We apply the Rosetta algorithm to repack the hydrophobic core of a β-peptide bundle while retaining both structure and stability.

We investigate the possible molecular mechanism of polyurethane esterase A, previously identified as responsible for degradation of a polyester polyurethane sample in Pseudomonas chlororaphis.

Human glucose transporters (GLUTs) facilitate the uptake of hexoses into cells. In cancer, the increased proliferation necessitates higher expression of GLUTs. This study demonstrates the inhibitory function of ganoderic acid A (GAA) on GLUT1/3.

A substrate binding induced conformational change was found to be essential for the occurrence of RrUGT3 catalyzed transglycosylation reactions.

ProtAgents is a de novo protein design platform based on multimodal LLMs, where distinct AI agents with expertise in knowledge retrieval, protein structure analysis, physics-based simulations, and results analysis tackle tasks in a dynamic setting.

. AlphaFold-based approaches for prediction of protein states and molecular dynamics simulations are integrated to characterize conformational ensembles and binding mechanisms of the SARS-CoV-2 spike Omicron variants with the host receptor ACE2.

The study investigates the molecular intricacies of SARS-CoV-2 RdRp via computational protein design, machine learning, and structural analyses, shedding light on mutational selection events impacting viral evolution and therapeutic strategies.

We employ peptide-binding molecularly imprinted nanoparticles (MINPs) to probe the regulatory conformations and scaffolding interactions governing Pyk2 kinase activation.

PIGNet2, a versatile protein–ligand interaction prediction model that performs well in both molecule identification and optimization, demonstrates its potential in early-stage drug discovery.

Peptide inhibitors against the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) are designed using a screening system for peptide-based inhibitors containing an α-helix region (SPICA) and structures predicted by AlphaFold2.

By using a genetically modified thermostable protein (KTQ5C), we have synthesized protein-stabilized goldnanoclusters (AuNC@KTQ5C) with advantageous properties, such as heat stable fluorescent emission and heat resistant peroxidase-like activity.

A computational strategy using synergetic energy and correlated configuration for redesigning enzymes (SECURE) is proposed for the thermostability engineering of multimeric proteins.

Utilizing a large protein language model, we have formulated a deep learning framework designed for predicting type II polyketide natural products.

Machine learning models provide an informed and efficient strategy to create novel peptide and protein sequences with the desired profiles.

We provide a family-wide assessment of accessible sites for covalent targeting that combined with AlphaFold revealed predicted small molecule binding pockets for guiding future inhibitor development of the DGK superfamily.

We propose a new deep learning DTA model 3DProtDTA, which utilises AlphaFold structure predictions in conjunction with the graph representation of proteins.

Alphafold2 was used to predict URAT1 protein structure, then the docking sites were identified, and three hit compounds were obtained through virtual screening and bioactivity verification.

AlphaFold2 predicts protein folds from sequence, which can be used for experimental structural biology, in construction and de novo protein design, prediction of complexes and perhaps even effects of mutations and conformational space exploration.

Yeast surface display using multi target selections enables monitoring of specificity profiles for thousands of proteins in parallel.

Since the last decade, various genome sequencing projects have led to the accumulation of an enormous set of genomic data; however, numerous protein-coding genes still need to be functionally characterized.

We evaluated the capabilities of ten molecular docking programs to predict the ligand binding poses (sampling power) and rank the binding affinities (scoring power).

Tethering ferredoxin (PetF) to photosystem I increased light-induced PetF-mediated electron transfer to soluble acceptors. Tethering was equivalent to using a ten-to-one molar ratio of soluble PetF to PSI.

We have described the π–π and cation–π interactions between the porphyrin ring and the protein part of porphyrin-containing proteins to better understand their stabilizing role.