Resumen de Long-read RNA-Seq: Quality control and benchmarking
Francisco José Pardo Palacios
This thesis presents the usage of long-read sequencing to overcome the limitations associated with conventional RNA-Seq, introducing significant innovations in this field. Long-read sequencing enables the capture of full-length transcripts and the detection of novel splicing variants, improving the accuracy of results compared to short-read sequencing, as there is no need for assembly, which could otherwise lead to chimeric isoforms.
As part of this work, the SQANTI3 tool has been designed and developed for the evaluation and filtering of transcriptomes. SQANTI3 classifies long-read transcription models into structural categories based on their splice junctions (SJ) and annotates a wide variety of quality features, such as the presence of non-canonical SJs or the reliability of Transcription Start and Termination Sites (TSS and TTS) detected using orthogonal data. It also includes an artifact filtering module based on machine learning or user-defined rules, as well as a "rescue" module to prevent the loss of complete genes due to excessive filtering. Finally, SQANTI3 integrates the functional annotation of transcriptomes with isoAnnot Lite, facilitating the analysis of isoform expression changes and their functional implications.
SQANTI3 was used in challenges 1 and 3 of the Long-read RNA-seq Genome Annotation Assessment Project (LRGASP), an international and multicenter effort to benchmark bioinformatic tools for long-read RNA-Seq data. Both challenges focused on the correct identification of transcripts in well-annotated organisms (challenge 1) and in non-model organisms with limited prior information (challenge 3). LRGASP provided participants with data from different sequencing technologies and protocols to submit the results obtained by their bioinformatics tools. These results were evaluated and compared using SQANTI3, highlighting the differences in transcriptomes obtained from the same sample depending on the data and methods used.
In summary, the work in thesis emphasizes the importance that long-read RNA-Seq can have in the future and how SQANTI3 is and will continue to be a key tool for the evaluation and improvement of transcriptome quality.
