Revolutionizing GPCR-ligand predictions: DeepGPCR with experimental validation for high-precision drug discovery.

G-protein coupled receptors (GPCRs), crucial in various diseases, are targeted of over 40% of approved drugs. However, the reliable acquisition of experimental GPCRs structures is hindered by their lipid-embedded conformations. Traditional protein-ligand interaction models falter in GPCR-drug interactions, caused by limited and low-quality structures. Generalized models, trained on soluble protein-ligand pairs, are also inadequate. To address these issues, we developed two models, DeepGPCR_BC for binary classification and DeepGPCR_RG for affinity prediction. These models use non-structural GPCR-ligand interaction data, leveraging graph convolutional networks and mol2vec techniques to represent binding pockets and ligands as graphs. This approach significantly speeds up predictions while preserving critical physical-chemical and spatial information. In independent tests, DeepGPCR_BC surpassed Autodock Vina and Schrödinger Dock with an area under the curve of 0.72, accuracy of 0.68 and true positive rate of 0.73, whereas DeepGPCR_RG demonstrated a Pearson correlation of 0.39 and root mean squared error of 1.34. We applied these models to screen drug candidates for GPR35 (Q9HC97), yielding promising results with three (F545-1970, K297-0698, S948-0241) out of eight candidates. Furthermore, we also successfully obtained six active inhibitors for GLP-1R. Our GPCR-specific models pave the way for efficient and accurate large-scale virtual screening, potentially revolutionizing drug discovery in the GPCR field.

Deep Learning-Based construction of a Drug-Like compound database and its application in virtual screening of HsDHODH inhibitors.

The process of virtual screening relies heavily on the databases, but it is disadvantageous to conduct virtual screening based on commercial databases with patent-protected compounds, high compound toxicity and side effects. Therefore, this paper utilizes generative recurrent neural networks (RNN) containing long short-term memory (LSTM) cells to learn the properties of drug compounds in the DrugBank, aiming to obtain a new and virtual screening compounds database with drug-like properties. Ultimately, a compounds database consisting of 26,316 compounds is obtained by this method. To evaluate the potential of this compounds database, a series of tests are performed, including chemical space, ADME properties, compound fragmentation, and synthesizability analysis. As a result, it is proved that the database is equipped with good drug-like properties and a relatively new backbone, its potential in virtual screening is further tested. Finally, a series of seedling compounds with completely new backbones are obtained through docking and binding free energy calculations.

A deep learning based multi-model approach for predicting drug-like chemical compound's toxicity.

Ensuring the safety and efficacy of chemical compounds is crucial in small-molecule drug development. In the later stages of drug development, toxic compounds pose a significant challenge, losing valuable resources and time. Early and accurate prediction of compound toxicity using deep learning models offers a promising solution to mitigate these risks during drug discovery. In this study, we present the development of several deep-learning models aimed at evaluating different types of compound toxicity, including acute toxicity, carcinogenicity, hERG_cardiotoxicity (the human ether-a-go-go related gene caused cardiotoxicity), hepatotoxicity, and mutagenicity. To address the inherent variations in data size, label type, and distribution across different types of toxicity, we employed diverse training strategies. Our first approach involved utilizing a graph convolutional network (GCN) regression model to predict acute toxicity, which achieved notable performance with Pearson R 0.76, 0.74, and 0.65 for intraperitoneal, intravenous, and oral administration routes, respectively. Furthermore, we trained multiple GCN binary classification models, each tailored to a specific type of toxicity. These models exhibited high area under the curve (AUC) scores, with an impressive AUC of 0.69, 0.77, 0.88, and 0.79 for predicting carcinogenicity, hERG_cardiotoxicity, mutagenicity, and hepatotoxicity, respectively. Additionally, we have used the approved drug dataset to determine the appropriate threshold value for the prediction score in model usage. We integrated these models into a virtual screening pipeline to assess their effectiveness in identifying potential low-toxicity drug candidates. Our findings indicate that this deep learning approach has the potential to significantly reduce the cost and risk associated with drug development by expediting the selection of compounds with low toxicity profiles. Therefore, the models developed in this study hold promise as critical tools for early drug candidate screening and selection.

In silico analysis of the wheat BBX gene family and identification of candidate genes for seed dormancy and germination.

BACKGROUND: B-box (BBX) proteins are a type of zinc finger proteins containing one or two B-box domains. They play important roles in development and diverse stress responses of plants, yet their roles in wheat remain unclear. RESULTS: In this study, 96 BBX genes were identified in the wheat genome and classified into five subfamilies. Subcellular localization prediction results showed that 68 TaBBXs were localized in the nucleus. Protein interaction prediction analysis indicated that interaction was one way that these proteins exerted their functions. Promoter analysis indicated that TaBBXs may play important roles in light signal, hormone, and stress responses. qRT-PCR analysis revealed that 14 TaBBXs were highly expressed in seeds compared with other tissues. These were probably involved in seed dormancy and germination, and their expression patterns were investigated during dormancy acquisition and release in the seeds of wheat varieties Jing 411 and Hongmangchun 21, showing significant differences in seed dormancy and germination phenotypes. Subcellular localization analysis confirmed that the three candidates TaBBX2-2 A, TaBBX4-2 A, and TaBBX11-2D were nuclear proteins. Transcriptional self-activation experiments further demonstrated that TaBBX4-2A was transcriptionally active, but TaBBX2-2A and TaBBX11-2D were not. Protein interaction analysis revealed that TaBBX2-2A, TaBBX4-2A, and TaBBX11-2D had no interaction with each other, while TaBBX2-2A and TaBBX11-2D interacted with each other, indicating that TaBBX4-2A may regulate seed dormancy and germination by transcriptional regulation, and TaBBX2-2A and TaBBX11-2D may regulate seed dormancy and germination by forming a homologous complex. CONCLUSIONS: In this study, the wheat BBX gene family was identified and characterized at the genomic level by bioinformatics analysis. These observations provide a theoretical basis for future studies on the functions of BBXs in wheat and other species.

Functional analysis of a wheat class III peroxidase gene, TaPer12-3A, in seed dormancy and germination.

BACKGROUND: Class III peroxidases (PODs) perform crucial functions in various developmental processes and responses to biotic and abiotic stresses. However, their roles in wheat seed dormancy (SD) and germination remain elusive. RESULTS: Here, we identified a wheat class III POD gene, named TaPer12-3A, based on transcriptome data and expression analysis. TaPer12-3A showed decreasing and increasing expression trends with SD acquisition and release, respectively. It was highly expressed in wheat seeds and localized in the endoplasmic reticulum and cytoplasm. Germination tests were performed using the transgenic Arabidopsis and rice lines as well as wheat mutant mutagenized with ethyl methane sulfonate (EMS) in Jing 411 (J411) background. These results indicated that TaPer12-3A negatively regulated SD and positively mediated germination. Further studies showed that TaPer12-3A maintained H2O2 homeostasis by scavenging excess H2O2 and participated in the biosynthesis and catabolism pathways of gibberellic acid and abscisic acid to regulate SD and germination. CONCLUSION: These findings not only provide new insights for future functional analysis of TaPer12-3A in regulating wheat SD and germination but also provide a target gene for breeding wheat varieties with high pre-harvest sprouting resistance by gene editing technology.

Generating and screening de novo compounds against given targets using ultrafast deep learning models as core components.

Deep learning is an artificial intelligence technique in which models express geometric transformations over multiple levels. This method has shown great promise in various fields, including drug development. The availability of public structure databases prompted the researchers to use generative artificial intelligence models to narrow down their search of the chemical space, a novel approach to chemogenomics and de novo drug development. In this study, we developed a strategy that combined an accelerated LSTM_Chem (long short-term memory for de novo compounds generation), dense fully convolutional neural network (DFCNN), and docking to generate a large number of de novo small molecular chemical compounds for given targets. To demonstrate its efficacy and applicability, six important targets that account for various human disorders were used as test examples. Moreover, using the M protease as a proof-of-concept example, we find that iteratively training with previously selected candidates can significantly increase the chance of obtaining novel compounds with higher and higher predicted binding affinities. In addition, we also check the potential benefit of obtaining reliable final de novo compounds with the help of MD simulation and metadynamics simulation. The generation of de novo compounds and the discovery of binders against various targets proposed here would be a practical and effective approach. Assessing the efficacy of these top de novo compounds with biochemical studies is promising to promote related drug development.

Identification of genetic loci and candidate genes underlying freezing tolerance in wheat seedlings.

KEY MESSAGE: Ten stable loci for freezing tolerance (FT) in wheat were detected by genome-wide association analysis. The putative candidate gene TaRPM1-7BL underlying the major locus QFT.ahau-7B.2 was identified and validated. Frost damage restricts wheat growth, development, and geographical distribution. However, the genetic mechanism of freezing tolerance (FT) remains unclear. Here, we evaluated FT phenotypes of 245 wheat varieties and lines, and genotyped them using a Wheat 90 K array. The association analysis showed that ten stable loci were significantly associated with FT (P < 1 × 10-4), and explained 6.45-26.33% of the phenotypic variation. In particular, the major locus QFT.ahau-7B.2 was consistently related to all nine sets of FT phenotypic data. Based on five cleaved amplified polymorphic sequence (CAPS) markers closely linked to QFT.ahau-7B.2, we narrowed down the target region to the 570.67-571.16 Mb interval (0.49 Mb) on chromosome 7B, in which four candidate genes were annotated. Of these, only TaRPM1-7BL exhibited consistent differential expression after low temperature treatment between freezing-tolerant and freezing-sensitive varieties. The results of cloning and whole-exome capture sequencing indicated that there were two main haplotypes for TaRPM1-7BL, including freezing-tolerant Hap1 and freezing-sensitive Hap2. Based on the representative SNP (+1956, A/G), leading to an amino acid change in the NBS domain, a CAPS marker (CAPS-TaRPM1-7BL) was developed and validated in 431 wheat varieties (including the above 245 materials) and 318 F2 lines derived from the cross of 'Annong 9267' (freezing-tolerant) × 'Yumai 9' (freezing-sensitive). Subsequently, the TaRPM1-7BL gene was silenced in 'Yumai 9' by virus-induced gene silencing (VIGS), and these silenced wheat seedlings exhibited enhanced FT phenotypes, suggesting that TaRPM1-7BL negatively regulates FT. These findings are valuable for understanding the complex genetic basis of FT in wheat.

A new method to estimate the histological stage of primary biliary cholangitis.

OBJECTIVE: To analyze the diagnostic efficacy of the periportal hypoechoic band (PHB) in the histological stage of patients with primary biliary cholangitis (PBC). METHODS: We prospectively included 77 cases of PBC pathologically or clinically confirmed, and high-frequency ultrasound (HFUS) measurements of the PHB were performed in all included patients. Ludwig staging system of histopathology was used as the gold standard. RESULTS: The width of the PHB was positively correlated with histological staging (r = 0.844, p < 0.001). By area under the receiving operating characteristic curve (AUROC), the best cutoff value for PHB for advanced stage (≥ stage 3) was 2.4 mm (AUROC: 0.934; 95%CI: 0.841-0.981) and 0.93 for sensitivity, and 0.91 for specificity, the concordance rates of PHB vs. liver biopsy was 90.3%. The correct rate for early-stage PBC was 87.9% and for the progressive stage was 93.1%. After multi-factor regression analysis, the PHB (OR = 1.331, CI = 1.105-1.603, p = 0.003) and total bilirubin (OR = 1.156, CI = 1.041-1.285, p = 0.007) were independent influencing factors for progressive PBC. CONCLUSIONS: Measurement of the PHB to assess advanced PBC is a simple and effective method. This method may complement current methods for the histological staging assessment of patients with PBC. REGISTRATION: Clinical trial registration: ChiCTR 2000032053, 2020/04/19. CLINICAL RELEVANCE STATEMENT: The measurement of periportal hypoechoic band (PHB) provides a simple and easy assessment of the degree of disease progression in patients with PBC and provides an important clinical reference in predicting the histological staging of PBC from an ultrasound perspective. KEY POINTS: • The PHB is correlated with histological staging in the patient with PBC. • The area under the ROC curves of PHB for detecting advanced stage (≥ stage 3) were 0.934 and 0.93 for sensitivity, and 0.91 for specificity, the concordance rates of PHB vs. liver biopsy was 90.3%. The application of PHB can better assess the advanced PBC. • Measurement of the PHB to assess advanced PBC is a simple and effective method that can significantly reduce the need for liver biopsy.

A New Co(II)-coordination Polymer: Fluorescence Performances, Loaded with Paclitaxel-hydrogel on Breast Cancer and Molecular Docking Study.

In the current context of the increasing incidence of breast cancer, we aim to develop an efficient drug carrier for breast cancer by constructing an innovative complex consisting of a metal-organic framework (MOF) and a hydrogel. The aim of this initiative is to provide new ideas and tools for breast cancer treatment strategies through scientific research, so as to address the current challenges in breast cancer treatment. In the present study, by employment of a new Co(II)-based coordination polymer with the chemical formula of [Co(H2O)(CH3OH)L]n (1) (H2L = 5-(1 H-tetrazol-5-yl)nicotinic acid) was solvothermally synthesized by reaction of Co(NO3)2·6H2O a mixed solvent of MeOH and water. The characteristics of ligand-based absorption and emission, as unveiled by ultraviolet and fluorescence spectroscopy tests, offer insights into the distinctive electronic transitions and structural features originating from the ligand in compound 1. Using natural polysaccharide hyaluronic acid (HA) and carboxymethyl chitosan (CMCS) as raw materials, HA/CMCS hydrogels were successfully prepared by chemical method and their internal morphology was studied by scanning electron microscopy. Using paclitaxel as a drug model, we further designed and synthesized a novel metal gel particle-loaded paclitaxel drug and evaluated its inhibitory effect on breast cancer cells. Finally, the hypothesized interactions between the complex and the receptor have been confirmed through molecular docking simulation, and multiple polar interactions have been verified, which further proves the potential anti-cancer capability and excellent bioactivity. Based on this, this composite material prepared from a novel Co(II)-coordinated polymer with paclitaxel hydrogel could provide a useful pathway for the identification and treatment of breast cancer.

Nuclear cGAS suppresses DNA repair and promotes tumorigenesis.

Accurate repair of DNA double-stranded breaks by homologous recombination preserves genome integrity and inhibits tumorigenesis. Cyclic GMP-AMP synthase (cGAS) is a cytosolic DNA sensor that activates innate immunity by initiating the STING-IRF3-type I IFN signalling cascade1,2. Recognition of ruptured micronuclei by cGAS links genome instability to the innate immune response3,4, but the potential involvement of cGAS in DNA repair remains unknown. Here we demonstrate that cGAS inhibits homologous recombination in mouse and human models. DNA damage induces nuclear translocation of cGAS in a manner that is dependent on importin-α, and the phosphorylation of cGAS at tyrosine 215-mediated by B-lymphoid tyrosine kinase-facilitates the cytosolic retention of cGAS. In the nucleus, cGAS is recruited to double-stranded breaks and interacts with PARP1 via poly(ADP-ribose). The cGAS-PARP1 interaction impedes the formation of the PARP1-Timeless complex, and thereby suppresses homologous recombination. We show that knockdown of cGAS suppresses DNA damage and inhibits tumour growth both in vitro and in vivo. We conclude that nuclear cGAS suppresses homologous-recombination-mediated repair and promotes tumour growth, and that cGAS therefore represents a potential target for cancer prevention and therapy.

Covalent and non-covalent interaction of myofibrillar protein and cyanidin-3-O-glucoside: focus on structure, binding sites and in vitro digestion properties.

BACKGROUND: The aim of this study was to investigate the effects of covalent and non-covalent interactions between myofibrillar protein (MP) and cyanidin-3-O-glucoside (C3G) on protein structure, binding sites, and digestion properties. Four methods of inducing covalent cross-linking were used in the preparation of MP-C3G conjugates, including tyrosinase-catalyzed oxidation, alkaline pH shift treatment, free radical grafting, and ultrasonic treatment. A comparison was made between MP-C3G conjugates and complexes, and the analysis included sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), C3G binding ratio, liquid chromatography-tandem mass spectrometry (LC-MS/MS), protein side-chain amino acids, circular dichroism spectroscopy, three-dimensional fluorescence, particle size, and in vitro simulated digestion. RESULTS: Covalent bonding between C3G and amino acid side chains in MP was confirmed by LC-MS/MS. In covalent bonding, tryptophan residues, free amino groups and sulfhydryl groups were all implicated. Among the 22 peptides covalently modified by C3G, 30 modification sites were identified, located in lysine, histidine, tryptophan, arginine and cysteine. In vitro simulated digestion experiments showed that the addition of C3G significantly reduced the digestibility of MP, with the covalent conjugate showing lower digestibility than the non-covalent conjugate. Moreover, the digestibility of protein decreased more during intestinal digestion, possibly because covalent cross-linking of C3G and MP further inhibited trypsin targeting sites (lysine and arginine). CONCLUSION: Covalent cross-linking of C3G with myofibrillar proteins significantly affected protein structure and reduced protein digestibility by occupying more trypsin binding sites. © 2023 Society of Chemical Industry.

Floral nectar reabsorption and a sugar concentration gradient in two long-spurred Habenaria species (Orchidaceae).

BACKGROUND: Floral nectar is the most common reward flowers offered to pollinators. The quality and quantity of nectar produced by a plant species provide a key to understanding its interactions with pollinators and predicting rates of reproductive success. However, nectar secretion is a dynamic process with a production period accompanied or followed by reabsorption and reabsorption remains an understudied topic. In this study, we compared nectar volume and sugar concentration in the flowers of two long-spurred orchid species, Habenaria limprichtii and H. davidii (Orchidaceae). We also compared sugar concentration gradients within their spurs and rates of reabsorption of water and sugars. RESULTS: Both species produced diluted nectar with sugar concentrations from 17 to 24%. Analyses of nectar production dynamics showed that as flowers of both species wilted almost all sugar was reabsorbed while the original water was retained in their spurs. We established a nectar sugar concentration gradient for both species, with differences in sugar concentrations at their spur's terminus and at their spur's entrance (sinus). Sugar concentration gradient levels were 1.1% in H. limprichtii and 2.8% in H. davidii, both decreasing as flowers aged. CONCLUSION: We provided evidence for the reabsorption of sugars but not water occurred in wilted flowers of both Habenaria species. Their sugar concentration gradients vanished as flowers aged suggesting a slow process of sugar diffusion from the nectary at the spur's terminus where the nectar gland is located. The processes of nectar secretion/reabsorption in conjunction with the dilution and hydration of sugar rewards for moth pollinators warrant further study.

Longitudinal association of remnant cholesterol with joint arteriosclerosis and atherosclerosis progression beyond LDL cholesterol.

BACKGROUND: Arteriosclerosis and atherosclerosis are closely related with cardiovascular disease (CVD) risk. Remnant cholesterol (RC) could predict CVD. However, its effect on joint arteriosclerosis and atherosclerosis progression remains unclear. This study aims to evaluate the association of RC with joint arteriosclerosis and atherosclerosis progression trajectories in the general population. METHODS: This study collected data across five biennial surveys of the Beijing Health Management Cohort from 2010 to 2019. Multi-trajectory model was used to determine the joint arteriosclerosis and atherosclerosis progression patterns by brachial-ankle pulse wave velocity (baPWV) and ankle brachial index (ABI). We also performed discordance analyses for RC vs. low density lipoprotein cholesterol (LDL-C) using ordinal logistics model. RESULTS: A total of 3186 participants were included, with three clusters following distinct arteriosclerosis and atherosclerosis progression patterns identified using a multi-trajectory model. In the multivariable-adjusted ordinal logistics analyses, RC was significantly associated with baPWV and ABI progression (OR: 1.20; 95% CI: 1.13-1.28, per 10 mg/dL). For the discordance analyses, the discordant low RC group was associated with decreased risk compared to the concordant group (OR: 0.73; 95% CI: 0.60-0.89). People with a high RC level were at an increased risk of joint arteriosclerosis and atherosclerosis progression, even with optimal LDL-C. CONCLUSIONS: RC is independently associated with joint arteriosclerosis and atherosclerosis progression beyond LDL-C. RC could be an earlier risk factor than LDL-C of arteriosclerosis and atherosclerosis in the general population.

Association of impaired sensitivity to thyroid hormones with hyperuricemia through obesity in the euthyroid population.

BACKGROUND: Impaired sensitivity to thyroid hormones is a newly proposed clinical entity associated with hyperuricemia in the subclinical hypothyroid population. However, it is unknown whether the association exists in the euthyroid population. This study aimed to explore the association of impaired sensitivity to thyroid hormones (assessed by the thyroid feedback quantile-based index [TFQI], parametric thyroid feedback quantile-based index [PTFQI], thyrotrophic thyroxine resistance index [TT4RI] and thyroid-stimulating hormone index [TSHI]) with hyperuricemia and quantify the mediating effect of body mass index BMI in the euthyroid population. METHODS: This cross-sectional study enrolled Chinese adults aged ≥ 20 years who participated in the Beijing Health Management Cohort (2008-2019). Adjusted logistic regression models were used to explore the association between indices of sensitivity to thyroid hormones and hyperuricemia. Odds ratios [OR] and absolute risk differences [ARD] were calculated. Mediation analyses were performed to estimate direct and indirect effects through BMI. RESULTS: Of 30,857 participants, 19,031 (61.7%) were male; the mean (SD) age was 47.3 (13.3) years; and 6,515 (21.1%) had hyperuricemia. After adjusting for confounders, individuals in the highest group of thyroid hormone sensitivity indices were associated with an increased prevalence of hyperuricemia compared with the lowest group (TFQI: OR = 1.18, 95% CI 1.04-1.35; PTFQI: OR = 1.20, 95% CI 1.05-1.36; TT4RI: OR = 1.17, 95% CI 1.08-1.27; TSHI: OR = 1.12, 95% CI 1.04-1.21). BMI significantly mediated 32.35%, 32.29%, 39.63%, and 37.68% of the associations of TFQI, PTFQI, TT4RI and TSHI with hyperuricemia, respectively. CONCLUSIONS: Our research revealed that BMI mediated the association between impaired sensitivity to thyroid hormones and hyperuricemia in the euthyroid population. These findings could provide useful evidence for understanding the interaction between impaired sensitivity to thyroid hormone and hyperuricemia in euthyroid individuals and suggest the clinical implications of weight control in terms of impaired thyroid hormones sensitivity.

Genome-wide meta-analysis identifies susceptibility loci for autoimmune hepatitis type 1.

BACKGROUND AND AIMS: Autoimmune hepatitis (AIH) is a rare and chronic autoimmune liver disease. While genetic factors are believed to play a crucial role in the etiopathogenesis of AIH, our understanding of these genetic risk factors is still limited. In this study, we aimed to identify susceptibility loci to further understand the pathogenesis of this disease. APPROACH AND RESULTS: We conducted a case-control association study of 1,622 Chinese patients with AIH type 1 and 10,466 population controls from two independent cohorts. A meta-analysis was performed to ascertain variants associated with AIH type 1. A single-nucleotide polymorphism within the human leukocyte antigen (HLA) region showed the strongest association with AIH (rs6932730: OR = 2.32; p = 9.21 × 10-73 ). The meta-analysis also identified two non-HLA loci significantly associated with AIH: CD28/CTLA4/ICOS on 2q33.3 (rs72929257: OR = 1.31; p = 2.92 × 10-9 ) and SYNPR on 3p14.2 (rs6809477: OR = 1.25; p = 5.48 × 10-9 ). In silico annotation, reporter gene assays, and CRISPR activation experiments identified a distal enhancer at 2q33.3 that regulated expression of CTLA4. In addition, variants near STAT1/STAT4 (rs11889341: OR = 1.24; p = 1.34 × 10-7 ), LINC00392 (rs9564997: OR = 0.81; p = 2.53 × 10-7 ), IRF8 (rs11117432: OR = 0.72; p = 6.10 × 10-6 ), and LILRA4/LILRA5 (rs11084330: OR = 0.65; p = 5.19 × 10-6 ) had suggestive association signals with AIH. CONCLUSIONS: Our study identifies two novel loci (CD28/CTLA4/ICOS and SYNPR) exceeding genome-wide significance and suggests four loci as potential risk factors. These findings highlight the importance of costimulatory signaling and neuro-immune interaction in the pathogenesis of AIH.

Deep learning-assisted LI-RADS grading and distinguishing hepatocellular carcinoma (HCC) from non-HCC based on multiphase CT: a two-center study.

OBJECTIVES: To develop a deep learning (DL) method that can determine the Liver Imaging Reporting and Data System (LI-RADS) grading of high-risk liver lesions and distinguish hepatocellular carcinoma (HCC) from non-HCC based on multiphase CT. METHODS: This retrospective study included 1049 patients with 1082 lesions from two independent hospitals that were pathologically confirmed as HCC or non-HCC. All patients underwent a four-phase CT imaging protocol. All lesions were graded (LR 4/5/M) by radiologists and divided into an internal (n = 886) and external cohort (n = 196) based on the examination date. In the internal cohort, Swin-Transformer based on different CT protocols were trained and tested for their ability to LI-RADS grading and distinguish HCC from non-HCC, and then validated in the external cohort. We further developed a combined model with the optimal protocol and clinical information for distinguishing HCC from non-HCC. RESULTS: In the test and external validation cohorts, the three-phase protocol without pre-contrast showed κ values of 0.6094 and 0.4845 for LI-RADS grading, and its accuracy was 0.8371 and 0.8061, while the accuracy of the radiologist was 0.8596 and 0.8622, respectively. The AUCs in distinguishing HCC from non-HCC were 0.865 and 0.715 in the test and external validation cohorts, while those of the combined model were 0.887 and 0.808. CONCLUSION: The Swin-Transformer based on three-phase CT protocol without pre-contrast could feasibly simplify LI-RADS grading and distinguish HCC from non-HCC. Furthermore, the DL model have the potential in accurately distinguishing HCC from non-HCC using imaging and highly characteristic clinical data as inputs. CLINICAL RELEVANCE STATEMENT: The application of deep learning model for multiphase CT has proven to improve the clinical applicability of the Liver Imaging Reporting and Data System and provide support to optimize the management of patients with liver diseases. KEY POINTS: • Deep learning (DL) simplifies LI-RADS grading and helps distinguish hepatocellular carcinoma (HCC) from non-HCC. • The Swin-Transformer based on the three-phase CT protocol without pre-contrast outperformed other CT protocols. • The Swin-Transformer provide help in distinguishing HCC from non-HCC by using CT and characteristic clinical information as inputs.

Deep Learning-Based Bioactive Therapeutic Peptide Generation and Screening.

Many bioactive peptides demonstrated therapeutic effects over complicated diseases, such as antiviral, antibacterial, anticancer, etc. It is possible to generate a large number of potentially bioactive peptides using deep learning in a manner analogous to the generation of de novo chemical compounds using the acquired bioactive peptides as a training set. Such generative techniques would be significant for drug development since peptides are much easier and cheaper to synthesize than compounds. Despite the limited availability of deep learning-based peptide-generating models, we have built an LSTM model (called LSTM_Pep) to generate de novo peptides and fine-tuned the model to generate de novo peptides with specific prospective therapeutic benefits. Remarkably, the Antimicrobial Peptide Database has been effectively utilized to generate various kinds of potential active de novo peptides. We proposed a pipeline for screening those generated peptides for a given target and used the main protease of SARS-COV-2 as a proof-of-concept. Moreover, we have developed a deep learning-based protein-peptide prediction model (DeepPep) for rapid screening of the generated peptides for the given targets. Together with the generating model, we have demonstrated that iteratively fine-tuning training, generating, and screening peptides for higher-predicted binding affinity peptides can be achieved. Our work sheds light on developing deep learning-based methods and pipelines to effectively generate and obtain bioactive peptides with a specific therapeutic effect and showcases how artificial intelligence can help discover de novo bioactive peptides that can bind to a particular target.

Ultrasound imaging findings in primary biliary cholangitis.

BACKGROUND: Our study aimed to analyze the characteristics of ultrasound images corresponding to each histological stage of primary biliary cholangitis (PBC). METHODS: We prospectively analyzed 75 confirmed cases of PBC and used liver biopsy as the gold standard to determine the disease stage. RESULTS: The typical ultrasound images of patients with PBC were characterized by a thickening of the portal vein wall (PVW) and periportal hypoechoic band (PHB) width with increasing histological stages, and significant increases in the left hepatic lobe diameter (LHLD) in stage II (by 64.0%) and stage III (by 69.2%). PHB width (r = 0.857, p < 0.001), PVW thickness (r = 0.488, p < 0.001), and spleen area (r = 0.8774, p < 0.001) were positively correlated with the histological stage. Significant changes were noted in the liver surface, echo texture, and edge between different stages. The areas under the receiver operating characteristic curve of composite indicators were 0.965 for predicting progressive PBC(≥ stage 2), and 0.926 for predicting advanced PBC(≥ stage 3). CONCLUSIONS: The ultrasound imaging characteristics of patients with PBC varied according to the histological staging. LHLD, PVW thickness, and PHB width were significantly correlated with the histological stage. A combination of high- and low-frequency ultrasound imaging can provide relevant cues regarding the degree of PBC progression and important clinical reference values. The application of all the ultrasound image findings as the composite indicators can better predict progressive and advanced PBC, providing important clinical reference values.

DeepBindBC: A practical deep learning method for identifying native-like protein-ligand complexes in virtual screening.

Identifying native-like protein-ligand complexes (PLCs) from an abundance of docking decoys is critical for large-scale virtual drug screening in early-stage drug discovery lead searching efforts. Providing reliable prediction is still a challenge for most current affinity predicting models because of a lack of non-binding data during model training, lost critical physical-chemical features, and difficulties in learning abstract information with limited neural layers. In this work, we proposed a deep learning model, DeepBindBC, for classifying putative ligands as binding or non-binding. Our model incorporates information on non-binding interactions, making it more suitable for real applications. ResNet model architecture and more detailed atom type representation guarantee implicit features can be learned more accurately. Here, we show that DeepBindBC outperforms Autodock Vina, Pafnucy, and DLSCORE for three DUD.E testing sets. Moreover, DeepBindBC identified a novel human pancreatic α-amylase binder validated by a fluorescence spectral experiment (Ka = 1.0 × 105 M). Furthermore, DeepBindBC can be used as a core component of a hybrid virtual screening pipeline that incorporating many other complementary methods, such as DFCNN, Autodock Vina docking, and pocket molecular dynamics simulation. Additionally, an online web server based on the model is available at http://cbblab.siat.ac.cn/DeepBindBC/index.php for the user's convenience. Our model and the web server provide alternative tools in the early steps of drug discovery by providing accurate identification of native-like PLCs.

Use of IDEAL-IQ in Quantifying Femoral Bone Marrow Involvement in Gaucher Disease.

OBJECTIVE: To quantitatively measure femoral bone marrow involvement in patients with Gaucher disease (GD) by using fat fraction (FF) derived from the iterative decomposition of water and fat with echo asymmetry and least-squares estimation quantitation (IDEAL-IQ) technique. METHODS: Bilateral femora of 23 patients with type 1 GD receiving low-dose imiglucerase treatment were prospectively scanned using structural magnetic resonance imaging sequences and an IDEAL-IQ sequence. Femoral bone marrow involvement was evaluated by both semiquantification (bone marrow burden [BMB] score based on magnetic resonance imaging structural images) and quantification (FF derived from IDEAL-IQ) methods. These patients were further divided into subgroups according to whether they underwent splenectomy or had bone complications. The interreader agreement of measurements and the correlation between FF and clinical status were statistically analyzed. RESULTS: In patients with GD, both BMB and FF evaluation of femora showed good interreader concordance (intraclass correlation coefficient = 0.98 and 0.99, respectively), and FF highly correlated with BMB score ( P < 0.001). The longer the duration of disease, the lower the FF ( P = 0.026). Femoral FF was lower in subgroups with splenectomy or bone complications than those without splenectomy or bone complications (0.47 ± 0.08 vs 0.60 ± 0.15, 0.51 ± 0.10 vs 0.61 ± 0.17, respectively, both P < 0.05). CONCLUSION: Femoral FF derived from IDEAL-IQ could be used to quantify femoral bone marrow involvement in patients with GD, and low bone marrow FF may predict worse outcomes of GD patients in this small-scale study.