smartSim: simulation of splice aware single cell smart-seq3 data.

Abstract

Motivation: Smart-seq3 is a powerful full-length single-cell RNA sequencing protocol that enables transcript-level quantification and splicing analysis by preserving unique molecular identifier (UMI) information. However, benchmarking computational tools for isoform reconstruction and splicing quantification remains challenging due to the lack of ground truth datasets. Herein, we present smartSim, a Smart-seq3 read simulator designed to generate realistic sequencing data that accurately reflects the complexities of single-cell transcriptomics.

Results: smartSim simulates known and novel splicing events, generates both UMI-containing and internal reads, and mimics protocol-specific biases by leveraging empirical data distributions. Our results show that smartSim-generated data closely resembles real Smart-seq3 datasets in terms of fragment length distributions, internal read counts, and read quality scores. It generates raw sequencing reads in FASTQ format, making it compatible with both genome- and transcriptome-based alignment tools. By extending simulation beyond gene-level quantification, smartSim provides a crucial resource for evaluating and improving computational methods for alternative splicing detection and isoform reconstruction in single-cell RNA sequencing.

Availability and implementation: smartSim is available at https://github.com/MarchalLab/smartSim.