Here we analyse genetic variation, population structure and diversity among 3,010 diverse Asian cultivated rice (Oryza sativa L.) genomes from the 3,000 Rice Genomes Project. Our results are consistent with the five major groups previously recognized, but also suggest several unreported subpopulations that correlate with geographic location. We identified 29 million single nucleotide polymorphisms, 2.4 million small indels and over 90,000 structural variations that contribute to within-and between-population variation. Using pan-genome analyses, we identified more than 10,000 novel full-length protein-coding genes and a high number of presence-absence variations. The complex patterns of introgression observed in domestication genes are consistent with multiple independent rice domestication events. The public availability of data from the 3,000 Rice Genomes Project provides a resource for rice genomics research and breeding.Asian cultivated rice is grown worldwide and comprises the staple food for half of the global population. It is envisaged that by the year 2035 1 feeding this growing population will necessitate that an additional 112 million metric tons of rice be produced on a smaller area of land, using less water and under more fluctuating climatic conditions, which will require that future rice cultivars be higher yielding and resilient to multiple abiotic and biotic stresses. The foundation of the continued improvement of rice cultivars is the rich genetic diversity within domesticated populations and wild relatives [2][3][4] . For over 2,000 years, two major types of O. sativa-O. sativa Xian group (here referred to as Xian/Indica (XI) and also known as , Hsien or Indica) and O. sativa Geng Group (here referred to as Geng/Japonica (GJ) and also known as , Keng or Japonica)-have historically been recognized [5][6][7] . Varied degrees of post-reproductive barriers exist between XI and GJ rice accessions 8 ; this differentiation between XI and GJ rice types and the presence of different varietal groups are well-documented at isozyme and DNA levels 6,9 . Two other distinct groups have also been recognized using molecular markers 10 ; one of these encompasses the Aus, Boro and Rayada ecotypes from Bangladesh and India (which we term the circum-Aus group (cA)) and the other comprises the famous Basmati and Sadri aromatic varieties (which we term the circum-Basmati group (cB)).Approximately 780,000 rice accessions are available in gene banks worldwide 11 . To enable the more efficient use of these accessions in future rice improvement, the Chinese Academy of Agricultural Sciences, BGI-Shenzhen and International Rice Research Institute sequenced over 3,000 rice genomes (3K-RG) as part of the 3,000 Rice Genomes Project 12. Here we present analyses of genetic variation in the 3K-RG that focus on important aspects of O. sativa diversity, single nucleotide polymorphisms (SNPs) and structural variation (deletions, duplications, inversions and translocations). We also construct a species pangenome consisting of 'core...

Heading date and photoperiod sensitivity are fundamental traits that determine rice adaptation to a wide range of geographic environments. By quantitative trait locus (QTL) mapping and candidate gene analysis using whole-genome re-sequencing, we found that Oryza sativa Pseudo-Response Regulator37 (OsPRR37; hereafter PRR37) is responsible for the Early heading7-2 (EH7-2)/Heading date2 (Hd2) QTL which was identified from a cross of late-heading rice 'Milyang23 (M23)' and early-heading rice 'H143'. H143 contains a missense mutation of an invariantly conserved amino acid in the CCT (CONSTANS, CO-like, and TOC1) domain of PRR37 protein. In the world rice collection, different types of nonfunctional PRR37 alleles were found in many European and Asian rice cultivars. Notably, the japonica varieties harboring nonfunctional alleles of both Ghd7/Hd4 and PRR37/Hd2 flower extremely early under natural long-day conditions, and are adapted to the northernmost regions of rice cultivation, up to 53° N latitude. Genetic analysis revealed that the effects of PRR37 and Ghd7 alleles on heading date are additive, and PRR37 down-regulates Hd3a expression to suppress flowering under long-day conditions. Our results demonstrate that natural variations in PRR37/Hd2 and Ghd7/Hd4 have contributed to the expansion of rice cultivation to temperate and cooler regions.

MYB-type transcription factors (TFs) play essential roles in plant growth, development and respond to environmental stresses. Role of MYB-related TFs of rice in drought stress tolerance is not well documented. Here, we report the isolation and characterization of a novel MYB-related TF, OsMYB48-1, of rice. Expression of OsMYB48-1 was strongly induced by polyethylene glycol (PEG), abscisic acid (ABA), H2O2, and dehydration, while being slightly induced by high salinity and cold treatment. The OsMYB48-1 protein was localized in the nucleus with transactivation activity at the C terminus. Overexpression of OsMYB48-1 in rice significantly improved tolerance to simulated drought and salinity stresses caused by mannitol, PEG, and NaCl, respectively, and drought stress was caused by drying the soil. In contrast to wild type plants, the overexpression lines exhibited reduced rate of water loss, lower malondialdehyde (MDA) content and higher proline content under stress conditions. Moreover, overexpression plants were hypersensitive to ABA at both germination and post-germination stages and accumulated more endogenous ABA under drought stress conditions. Further studies demonstrated that overexpression of OsMYB48-1 could regulate the expression of some ABA biosynthesis genes (OsNCED4, OsNCED5), early signaling genes (OsPP2C68, OSRK1) and late responsive genes (RAB21, OsLEA3, RAB16C and RAB16D) under drought stress conditions. Collectively, these results suggested that OsMYB48-1 functions as a novel MYB-related TF which plays a positive role in drought and salinity tolerance by regulating stress-induced ABA synthesis.

Electrocaloric cooling technologies, enabled by the discovery of the giant electrocaloric effect in dielectrics more than a decade ago, represents a zero-globalwarming-potential, environment-benign cooling alternative. Benefited from its nature as an electricity-driven capacitor, the electrocaloric working body renders the great advantages in the energy efficiency and the device integration. The decade-long efforts on advancing the technology revealed many promising material candidates with matured manufacturing protocols, as well as intriguing device prototypes for applications beyond the traditional vapor compression based cooling. This article presents the recent advances in electrocaloric cooling technologies, from material improvements to device demonstrations. The environmental impact and the energy efficiency of the technology were evaluated by the total effective warming impact and the material COP, respectively. In addition to the current progresses achieved by the decade-long research effort, the existing challenges and potential opportunities brought by the electrocaloric refrigeration will be discussed.

Automatic generation of paraphrases from a given sentence is an important yet challenging task in natural language processing (NLP). In this paper, we present a deep reinforcement learning approach to paraphrase generation. Specifically, we propose a new framework for the task, which consists of a generator and an evaluator, both of which are learned from data. The generator, built as a sequenceto-sequence learning model, can produce paraphrases given a sentence. The evaluator, constructed as a deep matching model, can judge whether two sentences are paraphrases of each other. The generator is first trained by deep learning and then further fine-tuned by reinforcement learning in which the reward is given by the evaluator. For the learning of the evaluator, we propose two methods based on supervised learning and inverse reinforcement learning respectively, depending on the type of available training data. Experimental results on two datasets demonstrate the proposed models (the generators) can produce more accurate paraphrases and outperform the stateof-the-art methods in paraphrase generation in both automatic evaluation and human evaluation.