Skip to main content
Fig. 2 | Genome Biology

Fig. 2

From: Performance difference of graph-based and alignment-based hybrid error correction methods for error-prone long reads

Fig. 2

Explanation of bimodal accuracy gain for graph-based method; model fitness and accuracy gain on real dataset. a Proportion of long reads with different solid k-mer number. Without loss of generosity, the simulated long reads with length of 1 kb and error rate of 25% are taken as example. A long read is labeled as “high-gain long read” of the accuracy gain is larger than 12.5% (half of the value of error rate), and “low-gain long read” otherwise. b Distribution of the single solid k-mer locations on the high-gain and low-gain long reads. Only the long reads with one solid k-mer are considered. c Distribution of solid k-mer number on the long reads with different lengths. d Accuracy gain distribution at each error rate level for alignment-based method. e Proportion of long reads with solid k-mer detected. Due to the mixture of different long read lengths, an upper boundary and lower boundary is provided. f Accuracy gain distribution at each error rate level for graph-based method. g Length distribution of long reads on which graph-based method (labeled as DBG) has better, equal, or worse performance than the alignment-based method (labeled as ALN). The p value is calculated by Wilcoxon rank sum test

Back to article page