Long single-molecule reads can resolve the complexity of the Influenza virus composed of rare, closely related mutant variants

Abstract

Abstract. As a result of a high rate of mutations and recombination events, an RNA-virus exists as a heterogeneous "swarm" of mutant variants. The long read length offered by single-molecule sequencing technologies allows each mutant variant to be sequenced in a single pass. However, high error rate limits the ability to reconstruct heterogeneous viral population composed of rare, related mutant variants. In this paper, we present 2SNV, a method able to tolerate the high error-rate of the single-molecule protocol and reconstruct mutant variants. 2SNV uses linkage between single nucleotide variations to efficiently distinguish them from read errors. To benchmark the sensitivity of 2SNV, we performed a single-molecule sequencing experiment on a sample containing a titrated level of known viral mutant variants. Our method is able to accurately reconstruct clone with frequency of 0.2% and distinguish clones that differed in only two nucleotides distantly located on the genome. 2SNV outperforms existing methods for full-length viral mutant reconstruction. The open source implementation of 2SNV is freely available for download at

Accurate assembly of minority viral haplotypes from next-generation sequencing through efficient noise reduction

“…1: Finding linked SNV pairs. CliqueSNV uses pairs of linked SNVs which have been previously introduced for the 2SNV method [3]. Let the major (minor) allele at a given genomic position be the allele observed in the majority (minority) of reads covering this position.…”

“…is the largest probability of observing the 2-haplotypes (22) among these n reads given that the variant (22) does not exist. It has been shown in [3] that after Bonferroni correction to multiple testing the value of p should satisfy the following inequality…”

“…Results of these tools show that they were unable to reliably detect variants of frequency even higher than the error rate of sequencing. The 2SNV algorithm [2] accommodated errors in links and was the first such tool to be able to detect haplotypes with a frequency below the error rate correctly. The proposed CliqueSNV method keeps the basic idea of 2SNV linkage analysis but develops a novel approach for collecting multiple SNV's and inference of true haplotypes.…”

“…CliqueSNV was validated on simulated and experimental data and compared with Savage [5], PredictHaplo [30], aBayesQR [1] and 2SNV [2]. We benchmark the tools using the results of a PacBio sequencing experiment on a sample containing a titrated level of known Influenza A (IAV) viral variants, on similar data sets for experimental HIV-1 single-read and paired-end Illumina data and simulated Illumina HIV-1 and IAV data.…”

Towards reproducible, transparent, and systematic benchmarking of omics computational tools

“…In these cases, instead of obtaining the golden standard, one can design a mock community (often referred as a synthetic mock community) by combining in vitro titrated proportions of community elements. The most popular mock communities are prepared as mixtures of known microbial organisms 31,32 . When microbial organisms are closely related with similar sequences, such as intra-host RNA-virus populations, one should challenge computational methods with various frequency profiles and include closely related pairs 19,[32][33][34] .…”