Fig. 3

Benchmarking using raw sequencing data from artificially mixed dsDNA. A Schematic representation of the artificial dsDNA used. There are two types of artificial dsDNA: one is a 538 bp wild-type DNA sequence present in the mouse genome, and the other is a DNA sequence with an 85 bp deletion (indicated by a red dotted line), which is hypothesized to be a result of genome editing by the SpCas9-sgRNA complex. B Schematic of the sequencing and analysis of mixed artificial dsDNA used for benchmarking. The artificially synthesized wild-type DNA sequence and the 85 bp deletion DNA sequence were mixed at a certain ratio (% expected) and then subjected to Sanger sequencing. The sequencing data were analyzed using a TIDE analysis tool such as PtWAVE. The values derived from the analysis are the observed values (% observed). C Evaluation of large-deletion detection capabilities in various variable selection modes. The detection rates of the 85 bp deletion dsDNA mixed at various ratios are plotted on the horizontal axis, with the initially expected detection rates on the vertical axis presented as a scatter plot. The approximation line was drawn using the linear_model Linear Regression fit function from the scikit-learn module. The linear relationship was evaluated using the Coefficient of Determination (CoD), which has a maximum value of one and can take negative values. Correlations were assessed using Pearson's correlation coefficient (R), and the p value from the no-correlation test was noted. D Evaluation of model uncertainty for various variable selection modes. The Bayesian Information Criterion (BIC) for each analysis result plotted in C is presented as a box plot. The horizontal axis represents the different variable selection modes and the vertical axis represents the BIC. The Wilcoxon signed-rank test was conducted as a two-sided test, and the p value was noted. The Wilcoxon signed-rank test was performed using the stats.wilcoxon function from the SciPy module