Applications of Next-Generation Sequencing Technologies to Diagnostic Virology

Barzon, Luisa; Lavezzo, Enrico; Militello, Valentina; Toppo, Stefano; Palù, Giorgio

Besides the specific limits of the different platforms, other common issues should be taken into account and carefully considered. The first sources of problems are certainly chimerical sequences, point mutations and insertions/deletions which occur during reverse transcription, PCR amplification or sequencing itself. In addition, PCR amplification bias might impact the relative frequencies of viral variants. The process of "data cleaning" consists of three main steps: sequence filtering, alignment and error correction, for which a panel of methods has been proposed [84, 135, 136] . Briefly, the filtering phase removes the low-quality sequences from the dataset, while the error correction separates true variants from those due to experimental noise. This step is based on the idea that errors are randomly distributed with low frequency, while sequences with real mutations can be clustered and their abundance quantified. A cluster of reads presenting the same mutations represents a haplotype and the size of the cluster is the haplotype frequency. Global haplotypes are more difficult to be identified, since the reads must be assembled in larger contigs and a unique solution in aligning overlapping reads is not guaranteed. To this respect, the advantage of 454 platform for haplotype reconstruction studies is evident, thanks to its longer reads output.