scRNA-seq with Nanopore sequencing: benchmark of approaches based on hybrid sequencing
Résumé
Long-read sequencing has recently been incorporated into single-cell RNA sequencing studies. Compared to the standard Illumina short-read library that primarily captures abundance information, Nanopore sequencing has several advantages : it can produce reads of several kb [1], it provides more precise transcriptome mapping/assembly and detects complex isoform variants. In the 10xGenomics Nanopore data analysis, a key challenge is the relatively low sequencing accuracy (~ 95% per base) which makes it difficult to detect the cell barcodes and UMI information in each Nanopore read [1]. To deal with this issue, several bioinformatics approaches have been developed using hybrid sequencing to guide the allocation of Nanopore reads using Illumina data. We performed a benchmark analysis of the available Single-Cell long-read hybrid sequencing based pipelines: Sicelore [2], Snuupy [3], Scnapbar [4], and Flames [5]. We first compared the performance of each pipeline for barcode-UMI detection and assignment, the ability of the pipeline to handle a large set of data sequenced on a PromethION, and the biological results at the gene level compared with Illumina results. We then selected the best tools to compare biological results at the transcript level. Our results showed that Snuupy outperforms Sicelore, Flames and Scnapbar in terms of barcode-UMI assignment. The polyA-independant algorithm of Snuupy assigns around 35% more reads than Sicelore on a MinION dataset. This UMI count increase allows us to obtain biological results closer to Illumina results than sicelore at the gene level. Sicelore showed good performance processing a large set of data from PromethION where Snuupy could not achieve its process, suggesting it cannot be used without important modifications of its code. The results produced by Flames on the assignment part were not convincing but as it is designed for isoform detection, we are now testing its isoform detection module coupled with Sicelore and Snuupy. Our aim is to select the best features in each pipeline to ultimately develop an optimal, scalable and reproducible pipeline to characterize transcript isoforms in single-cell data, using hybrid sequencing. We also foresee that the announced improvement of Nanopore sequencing accuracy may leverage the need of Illumina sequencing. Acknowledgements The IBENS genomics core facility was supported by the France Génomique national infrastructure, funded as part of the “Investissements d'Avenir” program managed by the Agence Nationale de la Recherche (contract ANR-10- INBS-09). References 1. Full article: Single-cell transcriptomics in the context of long-read nanopore sequencing. https://www.tandfonline.com/doi/full/10.1080/13102818.2021.1988868. 2. High throughput error corrected Nanopore single cell transcriptome sequencing | Nature Communications. https://www.nature.com/articles/s41467-020-17800-6. 3. Long, Y. et al. FlsnRNA-seq: protoplasting-free full-length single-nucleus RNA profiling in plants. Genome Biol. 22, 66 (2021). 4. Wang, Q. et al. Single-cell transcriptome sequencing on the Nanopore platform with ScNapBar. RNA 27, 763–770 (2021). 5. Tian, L. et al. Comprehensive characterization of single-cell full-length isoforms in human and mouse with long- read sequencing. Genome Biol. 22, 310 (2021).
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |