Techniques For Structure Prediction Research Articles

Restoring protein glycosylation with GlycoShape.

Despite ground-breaking innovations in experimental structural biology and protein structure prediction techniques, capturing the structure of the glycans that functionalize proteins remains a challenge. Here we introduce GlycoShape ( https://glycoshape.org ), an open-access glycan structure database and toolbox designed to restore glycoproteins to their native and functional form in seconds. The GlycoShape database counts over 500 unique glycans so far, covering the human glycome and augmented by elements from a wide range of organisms, obtained from 1 ms of cumulative sampling from molecular dynamics simulations. These structures can be linked to proteins with a robust algorithm named Re-Glyco, directly compatible with structural data in open-access repositories, such as the Research Collaboratory for Structural Bioinformatics Protein Data Bank (RCSB PDB) and AlphaFold Protein Structure Database, or own. The quality, performance and broad applicability of GlycoShape is demonstrated by its ability to predict N-glycosylation occupancy, scoring a 93% agreement with experiment, based on screening all proteins in the PDB with a corresponding glycoproteomics profile, for a total of 4,259 N-glycosylation sequons.

Chainsaw: protein domain segmentation with fully convolutional neural networks.

Protein domains are fundamental units of protein structure and play a pivotal role in understanding folding, function, evolution, and design. The advent of accurate structure prediction techniques has resulted in an influx of new structural data, making the partitioning of these structures into domains essential for inferring evolutionary relationships and functional classification. This article presents Chainsaw, a supervised learning approach to domain parsing that achieves accuracy that surpasses current state-of-the-art methods. Chainsaw uses a fully convolutional neural network which is trained to predict the probability that each pair of residues is in the same domain. Domain predictions are then derived from these pairwise predictions using an algorithm that searches for the most likely assignment of residues to domains given the set of pairwise co-membership probabilities. Chainsaw matches CATH domain annotations in 78% of protein domains versus 72% for the next closest method. When predicting on AlphaFold models, expert human evaluators were twice as likely to prefer Chainsaw's predictions versus the next best method. github.com/JudeWells/Chainsaw.

Harnessing the 3D-Beacons Network: A Comprehensive Guide to Accessing and Displaying Protein Structure Data.

Recent advancements in protein structure determination and especially in protein structure prediction techniques have led to the availability of vast amounts of macromolecular structures. However, the accessibility and integration of these structures into scientific workflows are hindered by the lack of standardization among publicly available data resources. To address this issue, we introduced the 3D-Beacons Network, a unified platform that aims to establish a standardized framework for accessing and displaying protein structure data. In this article, we highlight the importance of standardized approaches for accessing protein structure data and showcase the capabilities of 3D-Beacons. We describe four protocols for finding and accessing macromolecular structures from various specialist data resources via 3D-Beacons. First, we describe three scenarios for programmatically accessing and retrieving data using the 3D-Beacons API. Next, we show how to perform sequence-based searches to find structures from model providers. Then, we demonstrate how to search for structures and fetch them directly into a workflow using JalView. Finally, we outline the process of facilitating access to data from providers interested in contributing their structures to the 3D-Beacons Network. © 2024 The Authors. Current Protocols published by Wiley Periodicals LLC. Basic Protocol 1: Programmatic access to the 3D-Beacons API Basic Protocol 2: Sequence-based search using the 3D-Beacons API Basic Protocol 3: Accessing macromolecules from 3D-Beacons with JalView Basic Protocol 4: Enhancing data accessibility through 3D-Beacons.

Response score-based protein structure analysis for cancer prediction aided by the Internet of Things

Medical diagnosis through prediction and analysis is par excellence in integrating modern technologies such as the Internet of Things (IoT). With the aid of such technologies, clinical assessments are eased with protracted computing. Specifically, cancer research through structure prediction and analysis is improved through human and machine interventions sustaining precision improvements. This article, therefore, introduces a Protein Structure Prediction Technique based on Three-Dimensional Sequence. This sequence is modeled using amino acids and their folds observed during the pre-initial cancer stages. The observed sequences and the inflammatory response score of the structure are used to predict the impact of cancer. In this process, ensemble learning is used to identify sequence and folding responses to improve inflammations. This score is correlated with the clinical data for structures and their folds independently for determining the structure changes. Such changes through different sequences are handled using repeated ensemble learning for matching and unmatching response scores. The introduced idea integrated with deep ensemble learning and IoT combination, notably employing stacking method for enhanced cancer prediction precision and interdisciplinary collaboration. The proposed technique improves prediction precision, data correlation, and change detection by 11.83%, 8.48%, and 13.23%, respectively. This technique reduces correlation time and complexity by 10.43% and 12.33%, respectively.

Assessment of the Performances of the Protein Modeling Techniques Participating in CASP15 Using a Structure-Based Functional Site Prediction Approach: ResiRole.

Model quality assessments via computational methods which entail comparisons of the modeled structures to the experimentally determined structures are essential in the field of protein structure prediction. The assessments provide means to benchmark the accuracies of the modeling techniques and to aid with their development. We previously described the ResiRole method to gauge model quality principally based on the preservation of the structural characteristics described in SeqFEATURE functional site prediction models. We apply ResiRole to benchmark modeling group performances in the Critical Assessment of Structure Prediction experiment, round 15. To gauge model quality, a normalized Predicted Functional site Similarity Score (PFSS) was calculated as the average of one minus the absolute values of the differences of the functional site prediction probabilities, as found for the experimental structures versus those found at the corresponding sites in the structure models. The average PFSS per modeling group (gPFSS) correlates with standard quality metrics, and can effectively be used to rank the accuracies of the groups. For the free modeling (FM) category, correlation coefficients of the Local Distance Difference Test (LDDT) and Global Distance Test-Total Score (GDT-TS) metrics with gPFSS were 0.98239 and 0.87691, respectively. An example finding for a specific group is that the gPFSS for EMBER3D was higher than expected based on the predictive relationship between gPFSS and LDDT. We infer the result is due to the use of constraints imprinted by function that are a part of the EMBER3D methodology. Also, we find functional site predictions that may guide further functional characterizations of the respective proteins. The gPFSS metric provides an effective means to assess and rank the performances of the structure prediction techniques according to their abilities to accurately recount the structural features at predicted functional sites.

Enhancing alphafold-multimer-based protein complex structure prediction with MULTICOM in CASP15

To enhance the AlphaFold-Multimer-based protein complex structure prediction, we developed a quaternary structure prediction system (MULTICOM) to improve the input fed to AlphaFold-Multimer and evaluate and refine its outputs. MULTICOM samples diverse multiple sequence alignments (MSAs) and templates for AlphaFold-Multimer to generate structural predictions by using both traditional sequence alignments and Foldseek-based structure alignments, ranks structural predictions through multiple complementary metrics, and refines the structural predictions via a Foldseek structure alignment-based refinement method. The MULTICOM system with different implementations was blindly tested in the assembly structure prediction in the 15th Critical Assessment of Techniques for Protein Structure Prediction (CASP15) in 2022 as both server and human predictors. MULTICOM_qa ranked 3rd among 26 CASP15 server predictors and MULTICOM_human ranked 7th among 87 CASP15 server and human predictors. The average TM-score of the first predictions submitted by MULTICOM_qa for CASP15 assembly targets is ~0.76, 5.3% higher than ~0.72 of the standard AlphaFold-Multimer. The average TM-score of the best of top 5 predictions submitted by MULTICOM_qa is ~0.80, about 8% higher than ~0.74 of the standard AlphaFold-Multimer. Moreover, the Foldseek Structure Alignment-based Multimer structure Generation (FSAMG) method outperforms the widely used sequence alignment-based multimer structure generation.

Improving AlphaFold2-based protein tertiary structure prediction with MULTICOM in CASP15

Since the 14th Critical Assessment of Techniques for Protein Structure Prediction (CASP14), AlphaFold2 has become the standard method for protein tertiary structure prediction. One remaining challenge is to further improve its prediction. We developed a new version of the MULTICOM system to sample diverse multiple sequence alignments (MSAs) and structural templates to improve the input for AlphaFold2 to generate structural models. The models are then ranked by both the pairwise model similarity and AlphaFold2 self-reported model quality score. The top ranked models are refined by a novel structure alignment-based refinement method powered by Foldseek. Moreover, for a monomer target that is a subunit of a protein assembly (complex), MULTICOM integrates tertiary and quaternary structure predictions to account for tertiary structural changes induced by protein-protein interaction. The system participated in the tertiary structure prediction in 2022 CASP15 experiment. Our server predictor MULTICOM_refine ranked 3rd among 47 CASP15 server predictors and our human predictor MULTICOM ranked 7th among all 132 human and server predictors. The average GDT-TS score and TM-score of the first structural models that MULTICOM_refine predicted for 94 CASP15 domains are ~0.80 and ~0.92, 9.6% and 8.2% higher than ~0.73 and 0.85 of the standard AlphaFold2 predictor respectively.

Estimating protein complex model accuracy based on ultrafast shape recognition and deep learning in CASP15.

This article reports and analyzes the results of protein complex model accuracy estimation by our methods (DeepUMQA3 and GraphGPSM) in the 15th Critical Assessment of techniques for protein Structure Prediction (CASP15). The new deep learning-based multimeric complex model accuracy estimation methods are proposed based on the ensemble of three-level features coupling with deep residual/graph neural networks. For the input multimeric complex model, we describe it from three levels: overall complex features, intra-monomer features, and inter-monomer features. We designed an overall ultrafast shape recognition (USR) to characterize the relationship between local residues and the overall complex topology, and an inter-monomer USR to characterize the relationship between the residues of one monomer and the topology of other monomers. DeepUMQA3 (Group name: GuijunLab-RocketX) ranked first in the interface residue accuracy estimation of CASP15. The Pearson correlation between the interface residue Local Distance Difference Test (lDDT) predicted by DeepUMQA3 and the real lDDT is 0.570, the only method that exceeds 0.5. Among the top 5 methods, DeepUMQA3 achieved the highest Pearson correlation of lDDT on 25 out of 39 targets. GraphGPSM (Group name: GuijunLab-PAthreader) has TM-score Pearson correlations greater than 0.9 on 14 targets, showing a good ability to estimate the overall fold accuracy. The DeepUMQA3 server is available at http://zhanglab-bioinf.com/DeepUMQA/ and the GraphGPSM server is available at http://zhanglab-bioinf.com/GraphGPSM/.

BioLiP2: an updated structure database for biologically relevant ligand-protein interactions.

With the progress of structural biology, the Protein Data Bank (PDB) has witnessed rapid accumulation of experimentally solved protein structures. Since many structures are determined with purification and crystallization additives that are unrelated to a protein's in vivo function, it is nontrivial to identify the subset of protein-ligand interactions that are biologically relevant. We developed the BioLiP2 database (https://zhanggroup.org/BioLiP) to extract biologically relevant protein-ligand interactions from the PDB database. BioLiP2 assesses the functional relevance of the ligands by geometric rules and experimental literature validations. The ligand binding information is further enriched with other function annotations, including Enzyme Commission numbers, Gene Ontology terms, catalytic sites, and binding affinities collected from other databases and a manual literature survey. Compared to its predecessor BioLiP, BioLiP2 offers significantly greater coverage of nucleic acid-protein interactions, and interactions involving large complexes that are unavailable in PDB format. BioLiP2 also integrates cutting-edge structural alignment algorithms with state-of-the-art structure prediction techniques, which for the first time enables composite protein structure and sequence-based searching and significantly enhances the usefulness of the database in structure-based function annotations. With these new developments, BioLiP2 will continue to be an important and comprehensive database for docking, virtual screening, and structure-based protein function analyses.

A gated graph transformer for protein complex structure quality assessment and its performance in CASP15.

Proteins interact to form complexes to carry out essential biological functions. Computational methods such as AlphaFold-multimer have been developed to predict the quaternary structures of protein complexes. An important yet largely unsolved challenge in protein complex structure prediction is to accurately estimate the quality of predicted protein complex structures without any knowledge of the corresponding native structures. Such estimations can then be used to select high-quality predicted complex structures to facilitate biomedical research such as protein function analysis and drug discovery. In this work, we introduce a new gated neighborhood-modulating graph transformer to predict the quality of 3D protein complex structures. It incorporates node and edge gates within a graph transformer framework to control information flow during graph message passing. We trained, evaluated and tested the method (called DProQA) on newly-curated protein complex datasets before the 15th Critical Assessment of Techniques for Protein Structure Prediction (CASP15) and then blindly tested it in the 2022 CASP15 experiment. The method was ranked 3rd among the single-model quality assessment methods in CASP15 in terms of the ranking loss of TM-score on 36 complex targets. The rigorous internal and external experiments demonstrate that DProQA is effective in ranking protein complex structures. The source code, data, and pre-trained models are available at https://github.com/jianlin-cheng/DProQA.

Combining pairwise structural similarity and deep learning interface contact prediction to estimate protein complex model accuracy in CASP15.

Estimating the accuracy of quaternary structural models of protein complexes and assemblies (EMA) is important for predicting quaternary structures and applying them to studying protein function and interaction. The pairwise similarity between structural models is proven useful for estimating the quality of protein tertiary structural models, but it has been rarely applied to predicting the quality of quaternary structural models. Moreover, the pairwise similarity approach often fails when many structural models are of low quality and similar to each other. To address the gap, we developed a hybrid method (MULTICOM_qa) combining a pairwise similarity score (PSS) and an interface contact probability score (ICPS) based on the deep learning inter-chain contact prediction for estimating protein complex model accuracy. It blindly participated in the 15th Critical Assessment of Techniques for Protein Structure Prediction (CASP15) in 2022 and performed very well in estimating the global structure accuracy of assembly models. The average per-target correlation coefficient between the model quality scores predicted by MULTICOM_qa and the true quality scores of the models of CASP15 assembly targets is 0.66. The average per-target ranking loss in using the predicted quality scores to rank the models is 0.14. It was able to select good models for most targets. Moreover, several key factors (i.e., target difficulty, model sampling difficulty, skewness of model quality, and similarity between good/bad models) for EMA are identified and analyzed. The results demonstrate that combining the multi-model method (PSS) with the complementary single-model method (ICPS) is a promising approach to EMA.

Protein Prediction using Dictionary Based Regression Learning

Research Objectives: Molecular genetic data is managed by the information technology known as bioinformatics. Major concept involved in bioinformatics is a protein sequence. Amino acids bonded with peptide bond constitute the sequence of Protein and it is very essential to lead life. To predict sequence of amino acid, primary sequence obtains amino sequence folding and structures prediction. Research Novelty: In this manuscript, dictionary based regression learning and fuzzy genetic algorithm is proposed for protein prediction from structural analysis (DRL-FGA-PD-SA). In this input data are taken from Kaggle domain dataset. The extraction of protein features from given data is made through Kernel Matrix (KM) which extracts composition of amino acids, composition of dipeptide, composition of pseudo-amino-acid, composition of functional domain and distance-based features. Then fuzzy based genetic algorithm (FGA) update the selected features for classification of protein and the features are clustered. Finally, dictionary based regression learning (DRL) predicts the class of protein with conversion of values either 0’s or 1’s. Research Conclusions: The proposed method is executed on MATLAB. Here evaluation metrics as sensitivity, precision, f-measure, specificity, accuracy and error rate are outlined. Then the performance of the proposed DRL-FGA-PD-SA method provides 22.08%, 24.03%, 34.76% higher accuracy, 23.34%, 26.45%, 34.44% higher precision, compared with the existing systems such assdeep learning methods in protein structure prediction (FFNN-RNN-PD-SA), deep learning technique for protein structure prediction and protein design (DNN-PD-SA) and improved protein structure prediction using potentials from deep learning (DNN-SGDA-PD-SA) respectively.

CaB2O2F4: A Novel [BOF2]∞-Based Structural Template with a Strong Second Harmonic Generation Response and Large Band Gap

Nonlinear optical (NLO) materials are significant for their crucial role in the development of modern optoelectronic and laser technology. Recently, fluorooxoborates have aroused the widespread attention for their tight relationship between the micro-properties of functional [BOxF4–x](x+1)– (x = 0, 1, 2, and 3) modules and macroscopic performances. Nevertheless, whether fluorooxoborates with the anionic groups only composed of condensed [BO2F2] units exhibit a strong second harmonic generation (SHG) response remains unclear. Here, eight possible synthetic structures with the [BOF2]∞ chain in the CaB2O2F4 system were predicted by using evolutionary algorithm crystal structure prediction techniques. Among them, CaB2O2F4-I, II, IV, and V are centrosymmetric and CaB2O2F4-III, VI, VII, and VIII are non-centrosymmetric structures. Hence, CaB2O2F4-VI (C2) possesses an extremely short ultraviolet cutoff edge (147 nm) and the largest SHG coefficient among reported non-centrosymmetric fluorooxoborates only consisting of the [BO2F2] group. Also, the origin of the SHG coefficient was analyzed in detail through theoretical calculation.

Monoclinic C12: A new superhard carbon allotrope

Superhard materials have always been the focus of scientific research. Based on crystal structure prediction technique along with first-principles calculations, a new superhard carbon structure C12 (space group: C2/m) was proposed. This C12 structure is composed of sp2–sp3 hybrid bonding network. Analyses of its mechanical properties show that the structure is mechanically stable and exhibits strong anisotropy. The dynamic stability of C12 at ambient pressure is confirmed by phonon spectra. The analysis of its electronic band structure reveals that C12 is a narrow bandgap semiconductor. Its hardness reaches approximately 67 GPa, which means C12 is a potential superhard semiconductor material. C12 is expected to be used in new semiconductors in the future.

3D-equivariant graph neural networks for protein model quality assessment.

Quality assessment (QA) of predicted protein tertiary structure models plays an important role in ranking and using them. With the recent development of deep learning end-to-end protein structure prediction techniques for generating highly confident tertiary structures for most proteins, it is important to explore corresponding QA strategies to evaluate and select the structural models predicted by them since these models have better quality and different properties than the models predicted by traditional tertiary structure prediction methods. We develop EnQA, a novel graph-based 3D-equivariant neural network method that is equivariant to rotation and translation of 3D objects to estimate the accuracy of protein structural models by leveraging the structural features acquired from the state-of-the-art tertiary structure prediction method-AlphaFold2. We train and test the method on both traditional model datasets (e.g. the datasets of the Critical Assessment of Techniques for Protein Structure Prediction) and a new dataset of high-quality structural models predicted only by AlphaFold2 for the proteins whose experimental structures were released recently. Our approach achieves state-of-the-art performance on protein structural models predicted by both traditional protein structure prediction methods and the latest end-to-end deep learning method-AlphaFold2. It performs even better than the model QA scores provided by AlphaFold2 itself. The results illustrate that the 3D-equivariant graph neural network is a promising approach to the evaluation of protein structural models. Integrating AlphaFold2 features with other complementary sequence and structural features is important for improving protein model QA. The source code is available at https://github.com/BioinfoMachineLearning/EnQA. Supplementary data are available at Bioinformatics online.

Molecular dynamics study of tropical calcific pancreatitis (TCP) associated calcium-sensing receptor single nucleotide variation.

Tropical Calcific Pancreatitis (TCP) is a chronic non-alcoholic pancreatitis characterised by extensive calcification. The disease usually appears at a younger age and is more common in tropical regions. This disease’s progression can lead to pancreatic diabetes, which can subsequently lead to pancreatic cancer. The CASR gene encodes a calcium-sensing receptor (CaSR), which is a GPCR protein of class C. It is expressed in the islets of Langerhans, the parathyroid gland, and other tissues. It primarily detects small gradients in circulating calcium concentrations and couples this information to intracellular signalling, which helps to regulate PTH (parathyroid hormone) secretion and mineral ion homeostasis. From co-leading insulin release, CaSR modulates ductal HCO3− secretion, Ca2+ concentration, cell-cell communication, β-cell proliferation, and intracellular Ca2+ release. In pancreatic cancer, the CaSR limits cell proliferation. TCP-related four novel missense mutations P163R, I427S, D433H and V477A, found in CaSR extracellular domain (ECD) protein, which were reported in the mutTCPdb Database (https://lms.snu.edu.in/mutTCPDB/index.php). P163R mutation occurs in ligand-binding domain 1 (LBD-1) of the CaSR ECD. To investigate the influence of these variations on protein function and structural activity multiple in-silico prediction techniques such as SIFT, PolyPhen, CADD scores, and other methods have been utilized. A 500 ns molecular dynamic simulation was performed on the CaSR ECD crystal structure and the corresponding mutated models. Furthermore, Principal Component Analysis (PCA) and Essential Dynamics analysis were used to forecast collective motions, thermodynamic stabilities, and the critical subspace crucial to CaSR functions. The results of molecular dynamic simulations showed that the mutations P163R, I427S, D433H, and V477A caused conformational changes and decreased the stability of protein structures. This study also demonstrates the significance of TCP associated mutations. As a result of our findings, we hypothesised that the investigated mutations may have an effect on the protein’s structure and ability to interact with other molecules, which may be related to the protein’s functional impairment.

Design of novel deep-UV nonlinear optical materials with one-dimensional functional module [BO2]∞ chain and fluorine-driven short phase-matching

Design of deep-ultraviolet (deep-UV) nonlinear optical (NLO) materials is urgently needed owing to the lack of high-performance crystals to meet the contradictory criteria of deep-UV transparency, strong NLO response and moderate birefringence to reach deep-UV phase-matching. In this work, a “highly-polymerized units optimizing electron delocalization” strategy was proposed to explore novel types of deep-UV NLO anionic functional module and design new materials. Two novel NLO systems with new NLO anionic functional module-one dimensional [BO2]∞ chains were successfully designed. In total, four LiBO2 (I-IV) and five Li2BO2F (I–V) structures with good thermodynamically and dynamically stable/metastable were obtained by using evolutionary algorithm crystal structure prediction techniques. All these structures contain the [BO2]∞ infinite chains and have deep-UV transparency (< 177.3 nm). LiBO2 (II, IV) and Li2BO2F (I, IV-V) exhibit large NLO response (2.0–2.3 × KH2PO4) and large birefringence (0.093–0.146 @1064 nm). Remarkably, Li2BO2F–I possesses an extremely short phase-matching wavelength about 159 nm, which is beneficial from the introduction of fluorine to maintain the superior functional module [BO2]∞ and optimize the optical properties. The results indicate that [BO2]∞ can be as a new deep-UV NLO functional module, this work provides an insight into the design of new deep-UV compound with outstanding NLO performances.

High‐Pressure Crystal Structures of Pb2CO4 and PbC2O5 with Tetrahedral [CO4] and Pyrocarbonate [C2O5] atomic groups

AbstractThe search of the stable intermediate compounds in the system PbO−CO2 have been performed using crystal structure prediction technique in the pressure range of 0–50 GPa. Stable structures were found for the compositions Pb2CO4 and PbC2O5. The earlier known oxy‐carbonate Pb2CO4‐P212121 was shown to transform at 10 GPa into another oxy‐carbonate Pb2CO4‐Pnma and than at 22 GPa – into the sp3‐hybridized orthocarbonate with the same symmetry. PbC2O5 became thermodynamically stable at almost 10 GPa in the form of the pyrocarbonate structure P21/c. At 35–45 GPa, PbC2O5‐P21/c transforms into PbC2O5‐Fdd2 with the framework of [CO4] tetrahedra. Another energetically favorable, although dynamically unstable, pyrocarbonate structure PbC2O5‐P was found in the narrow pressure interval around 10 GPa. We suggest that some analogue of this structure can appear in experiment. One structure, sp3‐hybridized Pb2CO4‐Pnma, is dynamically and thermally stable at ambient pressure, and therefore can be decompressed and extracted from the high‐pressure environment.

DOCKGROUND membrane protein-protein set.

Membrane proteins are significantly underrepresented in Protein Data Bank despite their essential role in cellular mechanisms and the major progress in experimental protein structure determination. Thus, computational approaches are especially valuable in the case of membrane proteins and their assemblies. The main focus in developing structure prediction techniques has been on soluble proteins, in part due to much greater availability of the structural data. Currently, structure prediction of protein complexes (protein docking) is a well-developed field of study. However, the generic protein docking approaches are not optimal for the membrane proteins because of the differences in physicochemical environment and the spatial constraints imposed by the membranes. Thus, docking of the membrane proteins requires specialized computational methods. Development and benchmarking of the membrane protein docking approaches has to be based on high-quality sets of membrane protein complexes. In this study we present a new dataset of 456 non-redundant alpha helical binary interfaces. The set is significantly larger and more representative than the previously developed sets. In the future, it will become the basis for the development of docking and scoring benchmarks, similar to the ones for soluble proteins in the Dockground resource http://dockground.compbio.ku.edu.

Mimetic Neural Networks: A Unified Framework for Protein Design and Folding.

Recent advancements in machine learning techniques for protein structure prediction motivate better results in its inverse problem–protein design. In this work we introduce a new graph mimetic neural network, MimNet, and show that it is possible to build a reversible architecture that solves the structure and design problems in tandem, allowing to improve protein backbone design when the structure is better estimated. We use the ProteinNet data set and show that the state of the art results in protein design can be met and even improved, given recent architectures for protein folding.