Tea is the world's oldest and most popular caffeine-containing beverage with immense economic, medicinal, and cultural importance. Here, we present the first high-quality nucleotide sequence of the repeat-rich (80.9%), 3.02-Gb genome of the cultivated tea tree Camellia sinensis. We show that an extraordinarily large genome size of tea tree is resulted from the slow, steady, and long-term amplification of a few LTR retrotransposon families. In addition to a recent whole-genome duplication event, lineage-specific expansions of genes associated with flavonoid metabolic biosynthesis were discovered, which enhance catechin production, terpene enzyme activation, and stress tolerance, important features for tea flavor and adaptation. We demonstrate an independent and rapid evolution of the tea caffeine synthesis pathway relative to cacao and coffee. A comparative study among 25 Camellia species revealed that higher expression levels of most flavonoid- and caffeine- but not theanine-related genes contribute to the increased production of catechins and caffeine and thus enhance tea-processing suitability and tea quality. These novel findings pave the way for further metabolomic and functional genomic refinement of characteristic biosynthesis pathways and will help develop a more diversified set of tea flavors that would eventually satisfy and attract more tea drinkers worldwide.

This paper introduces DuReader, a new large-scale, open-domain Chinese machine reading comprehension (MRC) dataset, designed to address real-world MRC. DuReader has three advantages over previous MRC datasets: (1) data sources: questions and documents are based on Baidu Search and Baidu Zhidao 1 ; answers are manually generated.(2) question types: it provides rich annotations for more question types, especially yes-no and opinion questions, that leaves more opportunity for the research community. (3) scale: it contains 200K questions, 420K answers and 1M documents; it is the largest Chinese MRC dataset so far. Experiments show that human performance is well above current state-of-the-art baseline systems, leaving plenty of room for the community to make improvements. To help the community make these improvements, both DuReader 2 and baseline systems 3 have been posted online. We also organize a shared competition to encourage the exploration of more models. Since the release of the task, there are significant improvements over the baselines.

No abstract