Tea is a globally consumed non-alcohol beverage with great economic importance. However, lack of the reference genome has largely hampered the utilization of precious tea plant genetic resources towards breeding. To address this issue, we previously generated a high-quality reference genome of tea plant using Illumina and PacBio sequencing technology, which produced a total of 2,124 Gb short and 125 Gb long read data, respectively. A hybrid strategy was employed to assemble the tea genome that has been publicly released. We here described the data framework used to generate, annotate and validate the genome assembly. Besides, we re-predicted the protein-coding genes and annotated their putative functions using more comprehensive omics datasets with improved training models. We reassessed the assembly and annotation quality using the latest version of BUSCO. These data can be utilized to develop new methodologies/tools for better assembly of complex genomes, aid in finding of novel genes, variations and evolutionary clues associated with tea quality, thus help to breed new varieties with high yield and better quality in the future.