Telomere-to-telomere assembly by preserving contained reads

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Automated de novo genome assembly, particularly for diploid and polyploid genomes, faces challenges with graph simplification heuristics.
• String graph formulations commonly remove contained reads, which can inadvertently create assembly gaps.

Purpose of the Study:

• To mathematically derive the frequency of assembly gaps caused by contained read deletion.
• To develop a novel assembly algorithm to mitigate gaps introduced by contained reads.

Main Methods:

• Mathematical derivation of gap frequency related to germline and somatic heterozygous variants.
• Development of the Repeat-Aware Fragmenting Tool (RAFT) algorithm to fragment reads and create uniform length distributions.
• Empirical validation using simulated and real Oxford Nanopore Technologies (ONT) and Pacific Biosciences high-fidelity (PacBio HiFi) data.

Main Results:

• Assembly gaps due to contained read deletion are significantly more frequent in ONT reads compared to PacBio HiFi reads.
• Gap frequency decreases with increased sequencing depth.
• The RAFT algorithm demonstrated a significant reduction in assembly gaps on simulated datasets.
• Real data analysis showed RAFT achieved a twofold increase in contig NG50 and haplotype-resolved telomere-to-telomere (T2T) contigs compared to hifiasm.

Conclusions:

• Contained read deletion is a primary driver of gaps in genome assembly, with read length distribution being a key factor.
• RAFT effectively addresses the contained read problem, improving genome assembly quality and contiguity.
• The developed algorithm offers a significant advancement for telomere-to-telomere (T2T) genome assembly using long-read sequencing data.