Extended Data Fig. 1: Accurate HiFi string graph combining PacBio HiFi and ONT ultra-long reads.
From: Scalable telomere-to-telomere assembly for diploid and polyploid genomes with double graph

(a) Effect of contained reads in the string graph. Rectangles in orange and blue represent heterozygous HiFi reads from haplotype 1 and haplotype 2, respectively. Green rectangles are HiFi reads originating from homozygous regions, whereas red rectangles are contained reads. The string graph is constructed using all reads, except for two contained reads. (b) Hifiasm (UL) aligns ultra-long reads to the HiFi string graph with contained reads to alleviate the contained read problem. The alignment paths of ultra-long reads from haplotype 1 and haplotype 2 are represented by orange and blue lines, respectively. Despite being a contained read, h12 is retained as the critical read because it is covered by ultra-long reads u6 and u7. To ensure accurate graph cleaning, hifiasm (UL) also tracks the number of ultra-long reads that support each edge as its weight. For instance, the edge weight between h5 and h8 is 2 because ultra-long reads u4 and u5 cover it.