Fig. 3: Direct repeats contribute to the formation of high quality eccDNA.

a Depicts the correlation between the presence of direct and inverted repeats in the genesis of eccDNA. Created with BioRender.com. b Conducts a proportional analysis contrasting direct versus inverted repeats within eccDNAs extracted from assorted rice tissues against sequences generated at random. c Box-and-whisker plots compare direct and inverted repeat prevalence in eccDNAs and a Random Model (RM) across six biologically independent samples, with individual data points shown. Statistical significance for direct repeats is marked (p = 0.000349, adjusted p = 0.000699), not significant for inverted repeats (p = 0.058097), using a two-tailed Unpaired t test with df = 10. d Depicts the length distribution of direct repeats within eccDNAs. e Contrasts the lengths of DR-eccDNAs (n = 373, range: 100–4542 bp, median: 767 bp) against Other-eccDNAs (n = 25,345, range: 74–5196 bp, median: 414 bp), showing a notable difference (p < 0.0001). f Assesses the Circle Score quality of DR-eccDNAs in comparison to Other-eccDNAs, indicating superior quality in DR-eccDNAs (p < 0.0001). g Compares the GC content in DR-eccDNAs (n = 297, range: 25.87–75.16%, median: 45.77%) with Other-eccDNAs (n = 25,345, range: 17.09–83.27%, median: 55.38%), identifying significant variances (p < 0.0001). h Applies Hedges’ g to measure effect size in eccDNA comparisons. The p values for (e, f, g) were calculated using a two-sided Mann–Whitney test. Data representation includes mean values denoted by a “+“ symbol to indicate the central tendency, with box plots elucidating the distribution featuring minimum, maximum, median, 25th (Q1), and 75th (Q3) percentiles for detailed analysis. Statistical significance is highlighted by p values, with asterisks denoting significance levels (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Source data are provided as a Source Data file.