Application of a targeted-enrichment methodology for full-genome sequencing of Dengue 1-4, Chikungunya and Zika viruses directly from patient samples

Abstract

The frequency of epidemics caused by Dengue viruses 1-4, Zika virus and Chikungunya viruses have been on an upward trend in recent years driven primarily by uncontrolled urbanization, mobility of human populations and geographical spread of their shared vectors, Aedes aegypti and Aedes albopictus. Infections by these viruses present with similar clinical manifestations making them challenging to diagnose; this is especially difficult in regions of the world hyperendemic for these viruses. In this study, we present a targeted-enrichment methodology to simultaneously sequence the complete viral genomes for each of these viruses directly from clinical samples. Additionally, we have also developed a customized computational tool (BaitMaker) to design these enrichment baits. This methodology is robust in its ability to capture diverse sequences and is amenable to large-scale epidemiological studies. We have applied this methodology to two large cohorts: a febrile study based in Colombo, Sri Lanka taken during the 2009-2015 dengue epidemic (n = 170) and another taken during the 2016 outbreak of Zika virus in Singapore (n = 162). Results from these studies indicate that we were able to cover an average of 97.04% ± 0.67% of the full viral genome from samples in these cohorts. We also show detection of one DENV3/ZIKV co-infected patient where we recovered full genomes for both viruses.

Fig 1. Overview of BaitMaker.

BaitMaker has two modes, Conserved and Exhaustive mode to design 120 nucleotide (nt) baits for targeted genome enrichment methodology. (i) Conserved Mode: To generate conserved baits, complete or nearly complete viral genomes sequences were download from NCBI database. The mode generates baits for virus species at the region that is most conserved at species-level and present across the different strains. These conserved baits were identified by k-mer based search and clustering algorithm (PriMux) with the allowed number of six mismatches for bait hybridization. The baits were prioritized if the bait can target at least 70% of the sequences from NCBI. Assuming the sequencing library size is 300 nt, the overlapping baits within 500 nt were removed to get non-redundant conserved baits. For DENV, 65 conserved baits were generated. (ii) Exhaustive Mode: To generate exhaustive baits, viral sequences greater than 1000 nt were download from NCBI database. This mode generates all possible baits targeting all the sequences in the viral database and thus contains baits targeting all the different viral strains. Similar to the conserved mode, the overlapping baits and baits within 500 nt were removed. Then iteratively new baits were designed for the regions with no baits giving an exhaustive list of baits targeting all the genomic variation across different strains.

Fig 2. Genome coverage plots of unenriched and enriched samples of DENV1, ZIKV and CHIKV.

The top panel (A, C, E) are unenriched samples whereas the bottom panel (B, D, F) are matched enriched samples with baits. From the outermost circle, each plot reads as the viral genes in the genome, SNPs (single nucleotide polymorphisms) detected, depth of coverage at each position in log scale shown in red and the baits hybridizing to the genome with varying sequence identity (80–85% identity in blue, 85–90% in dark blue, 90–95% in green and 95–100% in dark green). The number within the circle indicates the percentage of sequencing reads mapped to the genome.

Fig 3. Effect of viral titers on the depth of coverage and breadth of genome covered in ZIKV dilution series.

Ct indicates ZIKV viral titers of the samples. The x-axis represents the depth of coverage in log scale and y-axis represents the corresponding percentage of genome covered at the given coverage depth. The depth of coverage at 90% genome covered is discussed in the main text.

Fig 4. Co-infection of DENV3 and ZIKV identified in one clinical subject by using the DENV, CHIKV and ZIKV baits panel.

From the outermost circle the plot reads as the DENV3 and ZIKV viral genomes, corresponding depth of coverage and the locations where the baits hybridize.