Fig. 3
From: BASE: a practical de novo assembler for large genomes using long NGS reads
17mer depth distribution of three human sequencing dataset. We count the depth of all 17mers in the sequencing reads, and calculate the frequency of each depth. About 35 % 17mers of YH 100 bp reads, 30 % 17mers of YH 150 bp reads and 53 % 17mers of NA12878D XTen reads having depth no more than 3. Then we say the NA12878D XTen reads should have more sequencing errors left after raw data filter and correction than YH dataset