Improving the sensitivity of long read overlap detection using grouped short k-mer matches

Abstract

Background Single-molecule, real-time sequencing (SMRT) developed by Pacific BioSciences produces longer reads than second-generation sequencing technologies such as Illumina. The increased read length enables PacBio sequencing to close gaps in genome assembly, reveal structural variations, and characterize the intra-species variations. It also holds the promise to decipher the community structure in complex microbial communities because long reads help metagenomic assembly. One key step in genome assembly using long reads is to quickly identify reads forming overlaps. Because PacBio data has higher sequencing error rate and lower coverage than popular short read sequencing technologies (such as Illumina), efficient detection of true overlaps requires specially designed algorithms. In particular, there is still a need to improve the sensitivity of detecting small overlaps or overlaps with high error rates in both reads. Addressing this need will enable better assembly for metagenomic data produced by third-generation sequencing technologies. Results In this work, we designed and implemented an overlap detection program named GroupK, for third-generation sequencing reads based on grouped k -mer hits. While using k -mer hits for detecting reads’ overlaps has been adopted by several existing programs, our method uses a group of short k -mer hits satisfying statistically derived distance constraints to increase the sensitivity of small overlap detection. Grouped k -mer hit was originally designed for homology search. We are the first to apply group hit for long read overlap detection. The experimental results of applying our pipeline to both simulated and real third-generation sequencing data showed that GroupK enables more sensitive overlap detection, especially for datasets of low sequencing coverage. Conclusions GroupK is best used for detecting small overlaps for third-generation sequencing data. It provides a useful supplementary tool to existing ones for more sensitive and accurate overlap detection. The source code is freely available at https://github.com/Strideradu/GroupK .

cPlot: Contig-Plotting Visualization for the Analysis of Short-Read Nucleotide Sequence Alignments

“…In this step, in order to intuitively analyze the similarity between the reference sequence and the query sequence, reads with a long length and close to the region of the reference sequence are optimally selected and aligned based on the region of the reference sequence. During the first basic alignment step, only reads with overlaps between the reference sequence and query sequence are aligned [ 18 ]. Read alignment results in the reads overlapping in a specific region of the reference sequence, and an additional analysis step is needed to accurately select the reads most similar to the reference sequence.…”

Hardware acceleration of genomics data analysis: challenges and opportunities

“…Third generation : Single Molecule Real-Time (SMRT) sequencing method commercialized by Pacific Biosciences, produces 100 - 200 gigabases per single 20-hour run, with approximately 30000 bp read lengths ( Du et al , 2019 ). Despite its high throughput, SMRT lacks the raw sequence accuracy of pyrosequencing at 87% compared to 99% ( Du et al , 2019 ). The cost per one million bases is $10 compared to approximately $2400 for pyrosequencing ( Liu et al , 2012 ).…”

Strobemers: an alternative to k-mers for sequence comparison

“…One approach is to use a small k-mer size and identify pairs (39) or groups (40) of them clustered tightly together, and it has been studied how to design the sampling distribution of seeds to optimize alignment sensitivity (41, 42). Multi-seed methods are robust to any mutation type and have shown to, e.g., improve overlap detection between long reads (43). However, they still match single k-mers individually and group them based on statistics after individual k-mer hits have been found.…”