Search

Abstract RNAs are fundamental in living cells and perform critical functions determined by their tertiary architectures. However, accurate modeling of 3D RNA structure remains a challenging problem. We present a novel method, DRfold, to predict RNA tertiary structures by simultaneous learning of local frame rotations and geometric restraints from experimentally solved RNA structures, where the learned knowledge is converted into a hybrid energy potential to guide RNA structure assembly. The method significantly outperforms previous approaches by >73.3% in TM-score on a sequence-nonredundant dataset containing recently released structures. Detailed analyses showed that the major contribution to the improvements arise from the deep end-to-end learning supervised with the atom coordinates and the composite energy function integrating complementary information from geometry restraints and end-to-end learning models. The open-source DRfold program with fast training protocol allows large-scale application of high-resolution RNA structure modeling and can be further improved with future expansion of RNA structure databases.

Predicting protein structure from sequence is still not possible for all proteins. Here, the authors introduce a method that integrates deep learning and information about protein co-evolution to guide the prediction of non-homologous protein structures with greater accuracy.

Despite the immense progress recently witnessed in protein structure prediction, the modeling accuracy for proteins that lack sequence and/or structure homologs remains to be improved. We developed an open-source program, DeepFold, which integrates spatial restraints predicted by multi-task deep residual neural-networks along with a knowledge-based energy function to guide its gradient-descent folding simulations. The results on large-scale benchmark tests showed that DeepFold creates full-length models with accuracy significantly beyond classical folding approaches and other leading deep learning methods. Of particular interest is the modeling performance on the most difficult targets with very few homologous sequences, where DeepFold achieved an average TM-score that was 40.3% higher than trRosetta and 44.9% higher than DMPfold. Furthermore, the folding simulations for DeepFold were 262 times faster than traditional fragment assembly simulations. These results demonstrate the power of accurately predicted deep learning potentials to improve both the accuracy and speed of ab initio protein structure prediction.

The development of effective and safe vaccines is the ultimate way to efficiently stop the ongoing COVID-19 pandemic, which is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Built on the fact that SARS-CoV-2 utilizes the association of its Spike (S) protein with the human angiotensin-converting enzyme 2 (ACE2) receptor to invade host cells, we computationally redesigned the S protein sequence to improve its immunogenicity and antigenicity. Toward this purpose, we extended an evolutionary protein design algorithm, EvoDesign, to create thousands of stable S protein variants that perturb the core protein sequence but keep the surface conformation and B cell epitopes. The T cell epitope content and similarity scores of the perturbed sequences were calculated and evaluated. Out of 22,914 designs with favorable stability energy, 301 candidates contained at least two pre-existing immunity-related epitopes and had promising immunogenic potential. The benchmark tests showed that, although the epitope restraints were not included in the scoring function of EvoDesign, the top S protein design successfully recovered 31 out of the 32 major histocompatibility complex (MHC)-II T cell promiscuous epitopes in the native S protein, where two epitopes were present in all seven human coronaviruses. Moreover, the newly designed S protein introduced nine new MHC-II T cell promiscuous epitopes that do not exist in the wildtype SARS-CoV-2. These results demonstrated a new and effective avenue to enhance a target protein’s immunogenicity using rational protein design, which could be applied for new vaccine design against COVID-19 and other pathogens.

Summary: Structure prediction for proteins lacking homologous templates in the Protein Data Bank (PDB) remains a significant unsolved problem. We developed a protocol, C-I-TASSER, to integrate interresidue contact maps from deep neural-network learning with the cutting-edge I-TASSER fragment assembly simulations. Large-scale benchmark tests showed that C-I-TASSER can fold more than twice the number of non-homologous proteins than the I-TASSER, which does not use contacts. When applied to a folding experiment on 8,266 unsolved Pfam families, C-I-TASSER successfully folded 4,162 domain families, including 504 folds that are not found in the PDB. Furthermore, it created correct folds for 85% of proteins in the SARS-CoV-2 genome, despite the quick mutation rate of the virus and sparse sequence profiles. The results demonstrated the critical importance of coupling whole-genome and metagenome-based evolutionary information with optimal structure assembly simulations for solving the problem of non-homologous protein structure prediction. Motivation: Taking advantage of the rapid progress in deep-learning technologies, residue-residue contact-map prediction recently achieved impressive breakthroughs. However, how to efficiently convert the binary contact maps into atomic-level structure models remains an important unsolved problem in ab initio protein structure prediction. In this work, we integrated the deep-learning contact-map predictions with cutting-edge threading assembly simulations and found that the inherent force field of the structural folding simulations is essential to maximize the potential of contact-assisted protein structure prediction, especially for the targets and regions that lack spatial restraints and sufficient evolutionary data.

Cerenkov Emission (CE) during external beam radiation therapy (EBRT) from a linear accelerator (Linac) has been demonstrated as a useful tool for radiotherapy quality assurance and potentially other applications for online tracking of tumors during treatment delivery. However, some of the current challenges that are impacting the potential of CE are related to the limited detection sensitivity and the lack of flexible tools to fit into an already complex treatment delivery environment. Silicon photomultiplier (SiPM) solid-state devices are new promising tools for low light detection due to their extreme sensitivity that mirrors photomultiplier tubes and yet have a form factor that is similar to silicon photodiodes, allowing for improved flexibility in device design that may help in the process of wider clinical applicability. In this paper, we assess the feasibility of using SiPMs to detect CE during EBRT from a Linac and contrast their performance with commercially available silicon photodiodes (PDs). We demonstrate the feasibility of the SiPM-based probes for standard dosimetry measurements. We also demonstrate that CE optical signals can be detected from tissue depths about five times greater than that for standard probes based on PDs, making our SiPM probe an enabling technology of CE measurements, particularly for deep tissue applications.

Accurate prediction of atomic-level protein structure is important for annotating the biological functions of protein molecules and for designing new compounds to regulate the functions. Template-based modeling (TBM), which aims to construct structural models by copying and refining the structural frameworks of other known proteins, remains the most accurate method for protein structure prediction. Due to the difficulty in recognizing distant-homology templates, however, the accuracy of TBM decreases rapidly when the evolutionary relationship between the query and template vanishes. In this study, we propose a new method, CEthreader, which first predicts residue-residue contacts by coupling evolutionary precision matrices with deep residual convolutional neural-networks. The predicted contact maps are then integrated with sequence profile alignments to recognize structural templates from the PDB. The method was tested on two independent benchmark sets consisting collectively of 1,153 non-homologous protein targets, where CEthreader detected 176% or 36% more correct templates with a TM-score >0.5 than the best state-of-the-art profile- or contact-based threading methods, respectively, for the Hard targets that lacked homologous templates. Moreover, CEthreader was able to identify 114% or 20% more correct templates with the same Fold as the query, after excluding structures from the same SCOPe Superfamily, than the best profile- or contact-based threading methods. Detailed analyses show that the major advantage of CEthreader lies in the efficient coupling of contact maps with profile alignments, which helps recognize global fold of protein structures when the homologous relationship between the query and template is weak. These results demonstrate an efficient new strategy to combine ab initio contact map prediction with profile alignments to significantly improve the accuracy of template-based structure prediction, especially for distant-homology proteins.

In this study, the major secretome components of Penicillium oxalicum 16 and Trichoderma reesei RUT-C30 under wheat bran (WB) and rice straw (RS) solid-state fermentation were systematically analyzed. The activities of the major components, e.g., cellulase, hemicellulase, and amylase, were consistent with their abundance in the secretomes. P. oxalicum 16 secreted more abundant glycoside hydrolases than T. reesei RUT-C30. The main up-regulated proteins from the induction of WB, compared with that from RS, were amylase, pectinase, and protease, whereas the main down-regulated enzymes were cellulase, hemicellulase, swollenin, and lytic polysaccharide monooxygenase (LPMO). Specifically, WB induced more β-1,4-glucosidases, namely, S8B0F3 (UniProt ID), and A0A024RWA5 than RS, but RS induced more β-1,4-exoglucanases and β-1,4-endoglucanases, namely, A0A024RXP8, A024SH76, S7B6D6, S7ZP52, A024SH20, A024S2H5, S8BGM3, S7ZX22, and S8AIJ2. The P. oxalicum 16 xylanases S8AH74 and S7ZA57 were the major components responsible for degrading soluble xylan, and S8BDN2 probably acted on solid-state hemicellulose instead of soluble xylan. The main hemicellulase component of T. reesei RUT-C30 in RS was the xyloglucanase A0A024S9Z6 with an abundance of 16%, but T. reesei RUT-C30 lacked the hemicellulase mannanase and had a small amount of the hemicellulase xylanase. P. oxalicum 16 produced more amylase than T. reesei RUT-C30, and the results suggest amylase S7Z6T2 may degrade soluble starch. The percentage of the glucoamylase S8B6D7 did not significantly change, and reached an average abundance of 5.5%. The major auxiliary degradation enzymes of P. oxalicum 16 were LPMOs S7Z716 and S7ZPW1, whereas those of T. reesei RUT-C30 were swollenin and LPMOs A0A024SM10, A0A024SFJ2, and A0A024RZP7.