We recently published the first draft genome of T. lutea obtained with the Illumina short-read technology. While this technology has a very low sequencing error rates, the assemblers are known to misassemble the long repeated sequences, resulting into the fragmentation of the genome assembly. The genome of T. lutea was re-sequenced with the long-read technology Pacific Bioscience. Indeed, long-read assemblers show efficiency to resolve the assembly of long repeated elements such as TEs. However, this technology have to date a high sequencing error rates and its combination with short-read Illumina data is became a common method to overcome this error rate. A de novo genome assembly was perform from the long-reads and was improved with Illumina short-read data, used in the first genome assembly version. The de novo genome of T. lutea is composed of 193 contigs and has a size of 82 Mb. A gain of around 30 Mb was obtained (+34%), compared to the previous genome assembly, having a size of 54 Mb and composed of 7,659 contigs. The size of the coding regions has fewly increased between the both genome versions. While the de novo genome assembly encodes for ~16,000 genes, corresponding to a coding region length of 28 Mb, the previous gene proportion of the draft genome version was of 25 Mb. This suggest that the new assembled regions are mostly repeated elements. This new genome version is by far away more accurate than the previous one and was suitable to properly detect and annotate the TE content. To identify potential autonomous TEs, we designed a pipeline named PiRATE (Pipeline to Retrieve and Annotate TEs) and conducted an accurate TE annotation in a de novo genome of T. lutea. We established that its genome is composed of 15.9% and 4.9% of Class I and Class II TEs respectively. Among them 3.8% and 15.95% correspond to potentially autonomous and non-autonomous TEs respectively.