Structure-based GC investigation shed new light on ITS2 evolution in Corydalis species

<strong> ITS2 was high-throughput sequenced from 29 </strong><em><strong>Corydalis</strong></em><strong> species. </strong> The quality of the raw data was surveyed and filtered by removing the low-quality bases and adapter sequences using the FastQC v0.11.8 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and Trimmomatic v 0.39 (Bolger, et al. 2014). Due to large inter-species differences of rDNA sequences and small intra-species differences of rDNA copies, we first employed GetOrganelle v.1.7.5 to assemble the main type of nrDNA (nuclear ribosomal DNA) in each species (Jin, et al. 2020), and then they were treated as reference sequence to extract nrDNA reads as much as possible by bowtie2, the sequencing depth was calculated by samtools v 1.9 (Wysoker, et al. 2009; Langmead and Salzberg 2012). Then, the extracted nrDNA reads were assembled by FLASH v1.2.11, which designed to extend the length of short reads by overlapping paired-end reads (Magoc and Salzberg 2011). Finally, BLAST v 2.7.1 was applied to obtain all ITS2/5.8S type from the assembled sequences, and MAFFT v7.407 and SeqKit v2.2.0 were used to calculate the number of each type (Camacho, et al. 2009; Shen, et al. 2016; Nakamura, et al. 2018). To eliminate the artificial types generated from sequencing and assembly, the type with only one assembled sequence was removed for downstream analyses.