CRISPR-Mediated Programmable 3D Genome Positioning and Nuclear Organization.

Programmable control of spatial genome organization is a powerful approach for studying how nuclear structure affects gene regulation and cellular function. Here, we develop a versatile CRISPR-genome organization (CRISPR-GO) system that can efficiently control the spatial positioning of genomic loci relative to specific nuclear compartments, including the nuclear periphery, Cajal bodies, and promyelocytic leukemia (PML) bodies. CRISPR-GO is chemically inducible and reversible, enabling interrogation of real-time dynamics of chromatin interactions with nuclear compartments in living cells. Inducible repositioning of genomic loci to the nuclear periphery allows for dissection of mitosis-dependent and -independent relocalization events and also for interrogation of the relationship between gene position and gene expression. CRISPR-GO mediates rapid de novo formation of Cajal bodies at desired chromatin loci and causes significant repression of endogenous gene expression over long distances (30-600 kb). The CRISPR-GO system offers a programmable platform to investigate large-scale spatial genome organization and function.

CRISPR/Cas9 in Genome Editing and Beyond.

The Cas9 protein (CRISPR-associated protein 9), derived from type II CRISPR (clustered regularly interspaced short palindromic repeats) bacterial immune systems, is emerging as a powerful tool for engineering the genome in diverse organisms. As an RNA-guided DNA endonuclease, Cas9 can be easily programmed to target new sites by altering its guide RNA sequence, and its development as a tool has made sequence-specific gene editing several magnitudes easier. The nuclease-deactivated form of Cas9 further provides a versatile RNA-guided DNA-targeting platform for regulating and imaging the genome, as well as for rewriting the epigenetic status, all in a sequence-specific manner. With all of these advances, we have just begun to explore the possible applications of Cas9 in biomedical research and therapeutics. In this review, we describe the current models of Cas9 function and the structural and biochemical studies that support it. We focus on the applications of Cas9 for genome editing, regulation, and imaging, discuss other possible applications and some technical considerations, and highlight the many advantages that CRISPR/Cas9 technology offers.

A One Precursor One siRNA Model for Pol IV-Dependent siRNA Biogenesis.

RNA-directed DNA methylation in Arabidopsis thaliana is driven by the plant-specific RNA Polymerase IV (Pol IV). It has been assumed that a Pol IV transcript can give rise to multiple 24-nt small interfering RNAs (siRNAs) that target DNA methylation. Here, we demonstrate that Pol IV-dependent RNAs (P4RNAs) from wild-type Arabidopsis are surprisingly short in length (30 to 40 nt) and mirror 24-nt siRNAs in distribution, abundance, strand bias, and 5'-adenine preference. P4RNAs exhibit transcription start sites similar to Pol II products and are featured with 5'-monophosphates and 3'-misincorporated nucleotides. The 3'-misincorporation preferentially occurs at methylated cytosines on the template DNA strand, suggesting a co-transcriptional feedback to siRNA biogenesis by DNA methylation to reinforce silencing locally. These results highlight an unusual mechanism of Pol IV transcription and suggest a "one precursor, one siRNA" model for the biogenesis of 24-nt siRNAs in Arabidopsis.

Interrogation of the dynamic properties of higher-order heterochromatin using CRISPR-dCas9.

Eukaryotic chromosomes feature large regions of compact, repressed heterochromatin hallmarked by Heterochromatin Protein 1 (HP1). HP1 proteins play multi-faceted roles in shaping heterochromatin, and in cells, HP1 tethering to individual gene promoters leads to epigenetic modifications and silencing. However, emergent properties of HP1 at supranucleosomal scales remain difficult to study in cells because of a lack of appropriate tools. Here, we develop CRISPR-engineered chromatin organization (EChO), combining live-cell CRISPR imaging with inducible large-scale recruitment of chromatin proteins to native genomic targets. We demonstrate that human HP1α tiled across kilobase-scale genomic DNA form novel contacts with natural heterochromatin, integrates two distantly targeted regions, and reversibly changes chromatin from a diffuse to compact state. The compact state exhibits delayed disassembly kinetics and represses transcription across over 600 kb. These findings support a polymer model of HP1α-mediated chromatin regulation and highlight the utility of CRISPR-EChO in studying supranucleosomal chromatin organization in living cells.

CRISPR-array-mediated imaging of non-repetitive and multiplex genomic loci in living cells.

Dynamic imaging of genomic loci is key for understanding gene regulation, but methods for imaging genomes, in particular non-repetitive DNAs, are limited. We developed CRISPRdelight, a DNA-labeling system based on endonuclease-deficient CRISPR-Cas12a (dCas12a), with an engineered CRISPR array to track DNA location and motion. CRISPRdelight enables robust imaging of all examined 12 non-repetitive genomic loci in different cell lines. We revealed the confined movement of the CCAT1 locus (chr8q24) at the nuclear periphery for repressed expression and active motion in the interior nucleus for transcription. We uncovered the selective repositioning of HSP gene loci to nuclear speckles, including a remarkable relocation of HSPH1 (chr13q12) for elevated transcription during stresses. Combining CRISPR-dCas12a and RNA aptamers allowed multiplex imaging of four types of satellite DNA loci with a single array, revealing their spatial proximity to the nucleolus-associated domain. CRISPRdelight is a user-friendly and robust system for imaging and tracking genomic dynamics and regulation.

Engineering 3D genome organization.

Cancers and developmental disorders are associated with alterations in the 3D genome architecture in space and time (the fourth dimension). Mammalian 3D genome organization is complex and dynamic and plays an essential role in regulating gene expression and cellular function. To study the causal relationship between genome function and its spatio-temporal organization in the nucleus, new technologies for engineering and manipulating the 3D organization of the genome have been developed. In particular, CRISPR-Cas technologies allow programmable manipulation at specific genomic loci, enabling unparalleled opportunities in this emerging field of 3D genome engineering. We review advances in mammalian 3D genome engineering with a focus on recent manipulative technologies using CRISPR-Cas and related technologies.

A Nucleosome Bridging Mechanism for Activation of a Maintenance DNA Methyltransferase.

DNA methylation and H3K9me are hallmarks of heterochromatin in plants and mammals, and are successfully maintained across generations. The biochemical and structural basis for this maintenance is poorly understood. The maintenance DNA methyltransferase from Zea mays, ZMET2, recognizes dimethylation of H3K9 via a chromodomain (CD) and a bromo adjacent homology (BAH) domain, which flank the catalytic domain. Here, we show that dinucleosomes are the preferred ZMET2 substrate, with DNA methylation preferentially targeted to linker DNA. Electron microscopy shows one ZMET2 molecule bridging two nucleosomes within a dinucleosome. We find that the CD stabilizes binding, whereas the BAH domain enables allosteric activation by the H3K9me mark. ZMET2 further couples recognition of H3K9me to an increase in the specificity for hemimethylated versus unmethylated DNA. We propose a model in which synergistic coupling between recognition of nucleosome spacing, H3K9 methylation, and DNA modification allows ZMET2 to maintain DNA methylation in heterochromatin with high fidelity.

The water lily genome and the early evolution of flowering plants.

Water lilies belong to the angiosperm order Nymphaeales. Amborellales, Nymphaeales and Austrobaileyales together form the so-called ANA-grade of angiosperms, which are extant representatives of lineages that diverged the earliest from the lineage leading to the extant mesangiosperms1-3. Here we report the 409-megabase genome sequence of the blue-petal water lily (Nymphaea colorata). Our phylogenomic analyses support Amborellales and Nymphaeales as successive sister lineages to all other extant angiosperms. The N. colorata genome and 19 other water lily transcriptomes reveal a Nymphaealean whole-genome duplication event, which is shared by Nymphaeaceae and possibly Cabombaceae. Among the genes retained from this whole-genome duplication are homologues of genes that regulate flowering transition and flower development. The broad expression of homologues of floral ABCE genes in N. colorata might support a similarly broadly active ancestral ABCE model of floral organ determination in early angiosperms. Water lilies have evolved attractive floral scents and colours, which are features shared with mesangiosperms, and we identified their putative biosynthetic genes in N. colorata. The chemical compounds and biosynthetic genes behind floral scents suggest that they have evolved in parallel to those in mesangiosperms. Because of its unique phylogenetic position, the N. colorata genome sheds light on the early evolution of angiosperms.

A genomic and epigenomic atlas of prostate cancer in Asian populations.

Prostate cancer is the second most common cancer in men worldwide1. Over the past decade, large-scale integrative genomics efforts have enhanced our understanding of this disease by characterizing its genetic and epigenetic landscape in thousands of patients2,3. However, most tumours profiled in these studies were obtained from patients from Western populations. Here we produced and analysed whole-genome, whole-transcriptome and DNA methylation data for 208 pairs of tumour tissue samples and matched healthy control tissue from Chinese patients with primary prostate cancer. Systematic comparison with published data from 2,554 prostate tumours revealed that the genomic alteration signatures in Chinese patients were markedly distinct from those of Western cohorts: specifically, 41% of tumours contained mutations in FOXA1 and 18% each had deletions in ZNF292 and CHD1. Alterations of the genome and epigenome were correlated and were predictive of disease phenotype and progression. Coding and noncoding mutations, as well as epimutations, converged on pathways that are important for prostate cancer, providing insights into this devastating disease. These discoveries underscore the importance of including population context in constructing comprehensive genomic maps for disease.

The evolution of hot Jupiters revealed by the age distribution of their host stars.

The unexpected discovery of hot Jupiters challenged the classical theory of planet formation inspired by our solar system. Until now, the origin and evolution of hot Jupiters are still uncertain. Determining their age distribution and temporal evolution can provide more clues into the mechanism of their formation and subsequent evolution. Using a sample of 383 giant planets around Sun-like stars collected from the kinematic catalogs of the Planets Across Space and Time project, we find that hot Jupiters are preferentially hosted by relatively younger stars in the Galactic thin disk. We subsequently find that the frequency of hot Jupiters declines with age as [Formula: see text]. In contrast, the frequency of warm/cold Jupiters shows no significant dependence on age. Such a trend is expected from the tidal evolution of hot Jupiters' orbits, and our result offers supporting evidence using a large sample. We also perform a joint analysis on the planet frequencies in the stellar age-metallicity plane. The result suggests that the frequencies of hot Jupiters and warm/cold Jupiters, after removing the age dependence are both correlated with stellar metallicities as [Formula: see text] and [Formula: see text], respectively. Moreover, we show that the above correlations can explain the bulk of the discrepancy in hot Jupiter frequencies inferred from the transit and radial velocity (RV) surveys, given that RV targets tend to be more metal-rich and younger than transits.

Haplotype-resolved genome assembly of Phanera championii reveals molecular mechanisms of flavonoid synthesis and adaptive evolution.

Phanera championii is a medicinal liana plant that has successfully adapted to hostile karst habitats. Despite extensive research on its medicinal components and pharmacological effects, the molecular mechanisms underlying the biosynthesis of critical flavonoids and its adaptation to karst habitats remain elusive. In this study, we performed high-coverage PacBio and Hi-C sequencing of P. championii, which revealed its high heterozygosity and phased the genome into two haplotypes: Hap1 (384.60 Mb) and Hap2 (383.70 Mb), encompassing a total of 58 612 annotated genes. Comparative genomes analysis revealed that P. championii experienced two whole-genome duplications (WGDs), with approximately 59.59% of genes originating from WGD events, thereby providing a valuable genetic resource for P. championii. Moreover, we identified a total of 112 genes that were strongly positively selected. Additionally, about 81.60 Mb of structural variations between the two haplotypes. The allele-specific expression patterns suggested that the dominant effect of P. championii was the elimination of deleterious mutations and the promotion of beneficial mutations to enhance fitness. Moreover, our transcriptome and metabolome analysis revealed alleles in different tissues or different haplotypes collectively regulate the synthesis of flavonoid metabolites. In summary, our comprehensive study highlights the significance of genomic and morphological adaptation in the successful adaptation of P. championii to karst habitats. The high-quality phased genomes obtained in this study serve as invaluable genomic resources for various applications, including germplasm conservation, breeding, evolutionary studies, and elucidation of pathways governing key biological traits of P. championii.

Histone deacetylase 6 deficiency protects the liver against ischemia/reperfusion injury by activating PI3K/AKT/mTOR signaling.

Liver transplantation (LT) is the only effective method to treat end-stage liver disease. Hepatic ischemia-reperfusion injury (IRI) continues to limit the prognosis of patients receiving LT. Histone deacetylase 6 (HDAC6) is a unique HDAC member involved in inflammation and apoptosis. However, its role and mechanism in hepatic IRI have not yet been reported. We examined HDAC6 levels in liver tissue from LT patients, mice challenged with liver IRI, and hepatocytes subjected to hypoxia/reoxygenation (H/R). In addition, HDAC6 global-knockout (HDAC6-KO) mice, adeno-associated virus-mediated liver-specific HDAC6 overexpressing (HDAC6-LTG) mice, and their corresponding controls were used to construct hepatic IRI models. Hepatic histology, inflammatory responses, and apoptosis were detected to assess liver injury. The molecular mechanisms of HDAC6 in hepatic IRI were explored in vivo and in vitro. Moreover, the HDAC6-selective inhibitor tubastatin A was used to detect the therapeutic effect of HDAC6 on liver IRI. Together, our results showed that HDAC6 expression was significantly upregulated in liver tissue from LT patients, mice subjected to hepatic I/R surgery, and hepatocytes challenged by hypoxia/reoxygenation (H/R) treatment. Compared with control mice, HDAC6 deficiency mitigated liver IRI by inhibiting inflammatory responses and apoptosis, whereas HDAC6-LTG mice displayed the opposite phenotype. Further molecular experiments show that HDAC6 bound to and deacetylated AKT and HDAC6 deficiency improved liver IRI by activating PI3K/AKT/mTOR signaling. In conclusion, HDAC6 is a key mediator of hepatic IRI that functions to promote inflammation and apoptosis via PI3K/AKT/mTOR signaling. Targeting hepatic HDAC6 inhibition may be a promising approach to attenuate liver IRI.

Weakened Interfacial Hydrogen Bond Connectivity Drives Selective Photocatalytic Water Oxidation toward H2O2 at Water/Brookite-TiO2 Interface.

The formation of H2O2 through the two-electron photocatalytic water oxidation reaction (WOR) is significant but encounters the competition with the four-electron O2 evolution reaction. Recent studies showed a crystal-phase dependence in H2O2 selectivity, where high purity brookite TiO2 (b-TiO2) exhibits remarkable H2O2 selectivity in contrast to the common rutile phase TiO2 (r-TiO2). However, the origin of such a structure-induced selectivity preference remains elusive, primarily due to the complexities associated with the solid-liquid interface system and excited-state chemistry. Herein, we conducted a comprehensive investigation into the selectivity mechanism of WOR at the water/b-TiO2(210) and water/r-TiO2(110) interfaces, employing first-principles molecular dynamics simulations and microkinetic analyses. Intriguingly, our results reveal that the intrinsic catalytic ability of the b-TiO2(210) itself does not enhance H2O2 selectivity compared to r-TiO2(110). Instead, it is the weakened interfacial hydrogen bond connectivity, modulated by the herringbone-like local atomic structure of the b-TiO2(210) surface, that determines the selectivity. Specifically, this weakened H-bond connectivity (i.e., local low water density) at the interface, owing to the strong water adsorption and distinct adsorption orientation, can stabilize the OH• radical and inhibit its deprotonation, leading to an improved H2O2 selectivity. By contrast, the relatively strong interface H-bond connectivity established over r-TiO2(110) accelerates the deprotonation of OH•, with the OH• coverage being 3 orders of magnitude lower than at the water/b-TiO2(210) interface. This study quantitatively demonstrates that the local H-bond structure (water density) at the liquid/solid interface significantly influences photocatalytic selectivity, and this insight may offer a rational approach to enhance the H2O2 selectivity.

Anisotropic Charge Migration on Perovskite Oxysulfide for Boosting Photocatalytic Overall Water Splitting.

The synthesis of photocatalysts with both broad light absorption and efficient charge separation is significant for a high solar energy conversion, which still remains to be a challenge. Herein, a narrow-bandgap Y2Ti2O5S2 (YTOS) oxysulfide nanosheet coexposed with defined {101} and {001} facets synthesized by a flux-assisted solid-state reaction was revealed to display the character of an anisotropic charge migration. The selective photodeposition of cocatalysts demonstrated that the {101} and {001} surfaces of YTOS nanosheets were the reduction and oxidation regions during photocatalysis, respectively. Density functional theory (DFT) calculations indicated a band energy level difference between the {101} and {001} facets of YTOS, which contributes to the anisotropic charge migration between them. The exposed Ti atoms on the {101} surface and S atoms on the {001} surface were identified, respectively, as reducing and oxidizing centers of YTOS nanosheets. This anisotropic charge migration generated a built-in electric field between these two facets, quantified by spatially resolved surface photovoltage microscopy, the intensity of which was found to be highly correlated with photocatalytic H2 production activity of YTOS, especially exhibiting a high apparent quantum yield of 18.2% (420 nm) after on-site modification of a Pt@Au cocatalyst assisted by Na2S-Na2SO3 hole scavengers. In conjunction with an oxygen-production photocatalyst and a [Co(bpy)3]2+/3+ redox shuttle, the YTOS nanosheets achieved a solar-to-hydrogen conversion efficiency of 0.15% via a Z-scheme overall water splitting. Our work is the first to confirm anisotropic charge migration in a perovskite oxysulfide photocatalyst, which is crucial for enhancing charge separation and surface catalytic efficiency in this material.

Copper-Catalyzed Asymmetric Remote C(sp3)-H Alkylation of N-Fluorocarboxamides with Glycine Derivatives and Peptides.

Saturated hydrocarbon bonds are ubiquitous in organic molecules; to date, the selective functionalization of C(sp3)-H bonds continues to pose a notorious difficulty, thereby garnering significant attention from the synthetic chemistry community. During the past several decades, a wide array of powerful new methodologies has been developed to enantioselectively modify C(sp3)-H bonds that is successfully applied in asymmetric formation of diverse bonds, including C-C, C-N, and C-O bonds; nevertheless, the asymmetric C(sp3)-H alkylation is elusive and, therefore, far less explored. In this work, we report a direct and robust strategy to construct highly valuable enantioenriched unnatural α-amino acid (α-AA) cognates and peptides by a copper-catalyzed enantioselective remote C(sp3)-H alkylation of N-fluorocarboxamides and readily accessible glycine esters under ambient conditions. The key to success lies in the optically active Cu catalyst generated through the coordination of glycine derivatives to enantiopure bisphosphine/Cu(I) species, which is beneficial to the single electronic reduction of N-fluorocarboxamides and the subsequent stereodetermining alkylation. More importantly, all types (primary, secondary, tertiary, and even α-oxy) of δ-C(sp3)-H bonds could be site- and stereospecifically activated by the kinetically favored 1,5-hydrogen atom transfer (1,5-HAT) step.

LASS2 enhances chemosensitivity to cisplatin by inhibiting PP2A-mediated ß-catenin dephosphorylation in a subset of stem-like bladder cancer cells.

BACKGROUND: The benefits of first-line, cisplatin-based chemotherapy for muscle-invasive bladder cancer are limited due to intrinsic or acquired resistance to cisplatin. Increasing evidence has revealed the implication of cancer stem cells in the development of chemoresistance. However, the underlying molecular mechanisms remain to be elucidated. This study investigates the role of LASS2, a ceramide synthase, in regulating Wnt/ß-catenin signaling in a subset of stem-like bladder cancer cells and explores strategies to sensitize bladder cancer to cisplatin treatment. METHODS: Data from cohorts of our center and published datasets were used to evaluate the clinical characteristics of LASS2. Flow cytometry was used to sort and analyze bladder cancer stem cells (BCSCs). Tumor sphere formation, soft agar colony formation assay, EdU assay, apoptosis analysis, cell viability, and cisplatin sensitivity assay were used to investigate the functional roles of LASS2. Immunofluorescence, immunoblotting, coimmunoprecipitation, LC-MS, PCR array, luciferase reporter assays, pathway reporter array, chromatin immunoprecipitation, gain-of-function, and loss-of-function approaches were used to investigate the underlying mechanisms. Cell- and patient-derived xenograft models were used to investigate the effect of LASS2 overexpression and a combination of XAV939 on cisplatin sensitization and tumor growth. RESULTS: Patients with low expression of LASS2 have a poorer response to cisplatin-based chemotherapy. Loss of LASS2 confers a stem-like phenotype and contributes to cisplatin resistance. Overexpression of LASS2 results in inhibition of self-renewal ability of BCSCs and increased their sensitivity to cisplatin. Mechanistically, LASS2 inhibits PP2A activity and dissociates PP2A from ß-catenin, preventing the dephosphorylation of ß-catenin and leading to the accumulation of cytosolic phospho-ß-catenin, which decreases the transcription of the downstream genes ABCC2 and CD44 in BCSCs. Overexpression of LASS2 combined with a tankyrase inhibitor (XAV939) synergistically inhibits tumor growth and restores cisplatin sensitivity. CONCLUSIONS: Targeting the LASS2 and ß-catenin pathways may be an effective strategy to overcome cisplatin resistance and inhibit tumor growth in bladder cancer patients.

Superstructured Carbon with Enhanced Kinetics for Zinc-Air Battery and Self-Powered Overall Water Splitting.

The present study proposes a novel engineering concept for the customization of functionality and construction of superstructure to fabricate 2D monolayered N-doped carbon superstructure electrocatalysts decorated with Co single atoms or Co2P nanoparticles derived from 2D bimetallic ZnCo-ZIF superstructure precursors. The hierarchically porous carbon superstructure maximizes the exposure of accessible active sites, enhances electron/mass transport efficiency, and accelerates reaction kinetics simultaneously. Consequently, the Co single atoms embedded N-doped carbon superstructure (Co-NCS) exhibits remarkable catalytic activity toward oxygen reduction reaction, achieving a half-wave potential of 0.886 V versus RHE. Additionally, the Co2P nanoparticles embedded N-doped carbon superstructure (Co2P-NCS) demonstrates high activity for both oxygen evolution reaction and hydrogen evolution reaction, delivering low overpotentials of 292 mV at 10 mA cm-2 and 193 mV at 10 mA cm-2 respectively. Impressively, when employed in an assembled rechargeable Zn-air battery, the as-prepared 2D carbon superstructure electrocatalysts exhibit exceptional performance with a peak power density of 219 mW cm-2 and a minimal charge/discharge voltage gap of only 1.16 V at 100 mA cm-2. Moreover, the cell voltage required to drive an overall water-splitting electrolyzer at a current density of 10 mA cm-2 is merely 1.69 V using these catalysts as electrodes.

ZmPHR1 contributes to drought resistance by modulating phosphate homeostasis in maize.

As an essential macronutrient, phosphorus (P) is often a limiting nutrient because of its low availability and mobility in soils. Drought is a major environmental stress that reduces crop yield. How plants balance and combine P-starvation responses (PSRs) and drought resistance is unclear. In this study, we identified the transcription factor ZmPHR1 as a major regulator of PSRs that modulates phosphate (Pi) signaling and homeostasis. We found that maize zmphr1 mutants had reduced P concentration and were sensitive to Pi starvation, whereas ZmPHR1-OE lines displayed elevated Pi concentration and yields. In addition, 57% of PSR genes and nearly 70% of ZmPHR1-regulated PSR genes in leaves were transcriptionally responsive to drought. Under moderate and early drought conditions, the Pi concentration of maize decreased, and PSR genes were up-regulated before drought-responsive genes. The ZmPHR1-OE lines exhibited drought-resistant phenotypes and reduced stomatal apertures, whereas the opposite was true of the zmphr1 mutants. ZmPT7-OE lines and zmspx3 mutants, which had elevated Pi concentration, also exhibited drought resistance, but zmpt7 mutants were sensitive to drought. Our results suggest that ZmPHR1 plays a central role in integrating Pi and drought signals and that Pi homeostasis improves the ability of maize to combat drought.

BatchDTA: implicit batch alignment enhances deep learning-based drug-target affinity estimation.

Candidate compounds with high binding affinities toward a target protein are likely to be developed as drugs. Deep neural networks (DNNs) have attracted increasing attention for drug-target affinity (DTA) estimation owning to their efficiency. However, the negative impact of batch effects caused by measure metrics, system technologies and other assay information is seldom discussed when training a DNN model for DTA. Suffering from the data deviation caused by batch effects, the DNN models can only be trained on a small amount of 'clean' data. Thus, it is challenging for them to provide precise and consistent estimations. We design a batch-sensitive training framework, namely BatchDTA, to train the DNN models. BatchDTA implicitly aligns multiple batches toward the same protein through learning the orders of candidate compounds with respect to the batches, alleviating the impact of the batch effects on the DNN models. Extensive experiments demonstrate that BatchDTA facilitates four mainstream DNN models to enhance the ability and robustness on multiple DTA datasets (BindingDB, Davis and KIBA). The average concordance index of the DNN models achieves a relative improvement of 4.0%. The case study reveals that BatchDTA can successfully learn the ranking orders of the compounds from multiple batches. In addition, BatchDTA can also be applied to the fused data collected from multiple sources to achieve further improvement.

Integrating data distribution prior via Langevin dynamics for end-to-end MR reconstruction.

PURPOSE: To develop a novel deep learning-based method inheriting the advantages of data distribution prior and end-to-end training for accelerating MRI. METHODS: Langevin dynamics is used to formulate image reconstruction with data distribution before facilitate image reconstruction. The data distribution prior is learned implicitly through the end-to-end adversarial training to mitigate the hyper-parameter selection and shorten the testing time compared to traditional probabilistic reconstruction. By seamlessly integrating the deep equilibrium model, the iteration of Langevin dynamics culminates in convergence to a fix-point, ensuring the stability of the learned distribution. RESULTS: The feasibility of the proposed method is evaluated on the brain and knee datasets. Retrospective results with uniform and random masks show that the proposed method demonstrates superior performance both quantitatively and qualitatively than the state-of-the-art. CONCLUSION: The proposed method incorporating Langevin dynamics with end-to-end adversarial training facilitates efficient and robust reconstruction for MRI. Empirical evaluations conducted on brain and knee datasets compellingly demonstrate the superior performance of the proposed method in terms of artifact removing and detail preserving.