Single-cell transcriptomics combined with third-generation sequencing, along with the emerging field of spatial transcriptomics, produces new types of data. These new types of data demand specialized tools for quality control, pre-processing, as well as careful benchmarking. Short-read single-cell sequencing profiles gene expression but has limited ability to resolve the expressed isoforms of a gene, even though isoforms can have opposite functions. Long-read sequencing can potentially sequence full-length transcripts, when combined with single-cell protocols, it resolves isoform counts and single-nucleotide variants at the single-cell level. Meanwhile, in the fast-growing spatial transcriptomics field, various protocols have been released, with different resolution and chemistry, requiring well-designed benchmarking datasets to evaluate them.
In this seminar I will present three pieces of my PhD work to address these issues. I will start with seqsizzle, an open-source long-read quality checking tool for visualizing and summarizing read structure. The second is FLAMESv2, a flexible, R-based pipeline that processes long-read single-cell or spatial data from raw reads through to analysis, with a wide range of protocol support. Finally, I will present a benchmarking study of spatial transcriptomics in mouse spleens after malaria challenge. Together, these tools and benchmarks help researchers analyze data from new transcriptomics technologies and choose the appropriate technologies.