Next generation sequencing and bioinformatic approaches to identify dengue viruses from non-invasive clinical samples

Next-generation sequencing (NGS) is becoming feasible for identification or diagnosis of human infectious diseases. In the present study, we aimed to develop a bioinformatic method for dengue virus identification by comparison against different database for variant types of samples and BLASTn and ViPR databases. We collected 12 different types of samples from five dengue-virus-infected patients and one sample from a healthy volunteer (control). RNA libraries were sequenced from all sample filtrates and enriched in virus load. The query sequences were aligned against NCBI BLASTn and ViPR databases. The fraction of double-stranded DNA query sequences against BLASTn was >75% compared with those against ViPR, and the abundance of single-stranded RNA in libraries against ViPR was more than those from the BLASTn database. The fraction of ssRNA against ViPR was higher than that against BLASTn in serum samples compared with urine and saliva samples. The taxonomic classification of viral reads grouped in different families indicated all samples were classified in dsDNA Herpesvirales and Poxviridae families, and the sequencing reads grouped in the Flaviviridae family were mainly seen in serum and urine samples. Comparing the components of total sequencing reads from each sample and dengue virus sequences, human reads comprised 85% of the total reads from serum and saliva, and 25% of total virus counts from urine samples. Even though the number of read counts was higher in serum than urine and saliva samples, the sequencing classification may provide an effective primary clue to detect dengue virus or similar related symptomatic virus infection.