Fig. 1: Assessment of analysis pipelines in eccDNA identification.

a Schematic overview of the benchmarking workflow used to compare the performance of bioinformatic pipelines. The cell line, healthy tissue and tumor illustration were created in BioRender. Gao, X. (2024) BioRender.com/h74t202. ‘Std’ represents standard deviation. b Performance comparison of analysis pipelines at a simulated sequencing depth of 50X (bms, bwa-mem-samblaster; mIs, microDNA.InOne.sh). Data are presented as mean values +/- SEM. c Impact of simulated sequencing depth on eccDNA identification accuracy. Data are presented as mean values +/- SEM. d Impact of simulated sequencing depth on eccDNA identification duplication rates. Centre line, median; box limits, upper and lower quartiles; whiskers, 1.5x interquartile range. e Impact of chimeric DNA proportion on eccDNA identification recall. Data are presented as mean values +/- SEM. The ‘n’ in the figure represents the number of datasets successfully analyzed by corresponding analysis pipeline and is used as the sample size to evaluate the performance of the respective analysis pipeline in the analysis. For panels (b, c and d) n = 7, except for ecc_finder (asm-sr) with n = 6 and ECCsplorer with n = 6 when depth ≤ 10, n = 5 for depths between 15X and 25X, n = 4 for depths between 30X and 40X, and n = 3 for depths higher than 40X. For panel (e) n = 4 for only 4 datasets contain chimeric eccDNA. Source data are provided as a Source Data file.