# Plastid RNA Editing in Glycyrrhiza uralensis: Landscape Characterization and Comparative Assessment of RNA-Seq Library Strategies for Detection

## TL;DR

This study identifies 38 RNA editing sites in the medicinal plant Glycyrrhiza uralensis and compares RNA sequencing methods for detecting them.

## Contribution

The study provides the first comprehensive characterization of plastid RNA editing in G. uralensis and evaluates RNA-seq library strategies for detection.

## Key findings

- 38 high-confidence RNA editing sites were identified across 19 genes in G. uralensis.
- Total RNA-seq outperformed rRNA-depleted and mRNA-seq methods in detecting RNA editing sites.
- mRNA-seq can still recover many editing sites with stringent, strand-specific filtering.

## Abstract

Background: Plastid RNA editing is widespread in angiosperms yet remains underexplored in the medicinal non-model species Glycyrrhiza uralensis. This study aimed to (i) comprehensively identify plastid RNA editing sites in G. uralensis, and (ii) compare the detection performance of three library construction strategies: total RNA-seq, rRNA-depleted RNA-seq, and mRNA-seq. Methods: Leaf tissue was used from three wild-sampled individual plants. Plastomes were assembled with GetOrganelle v1.7.0 and annotated using PGA. Strand-specific RNA-seq libraries were mapped to sample-matched plastomes using HISAT2 v2.2.1. Variants were identified using REDItools v2.0 under uniform thresholds. Candidate sites were visually verified in IGV v2.12.3, and read origins were confirmed by BLAST v2.13.0+; artifacts were removed via strand-specific filtering. Results: After stringent filtering, 38 high-confidence RNA editing sites were identified across 19 genes. Total RNA seq performed the best, detecting 37/38 sites consistently, whereas rRNA-depleted libraries detected fewer genuine sites and produced numerous rRNA-linked, noncanonical, noncoding-strand-dominant artifacts. Despite very low rates of plastid mapping, mRNA seq recovered a large fraction of bona fide sites under stringent, strand-aware filtering. Conclusions: We establish a set of 38 high-confidence plastid RNA editing sites in G. uralensis and suggest potential adaptive implications of editing in ndh-related genes. Methodologically, total RNA-seq is recommended for identification using de novo RNA editing due to its high sensitivity and low false-positive rate; publicly available poly(A)-selected mRNA-seq datasets can be repurposed to reliably retrieve plastid RNA editing sites when stringent strand-specific filtering is applied.