Analysis of ultra-deep pyrosequencing and cloning based sequencing of the basic core promoter/precore/core region of...

The aims of this study were to develop bioinformatics tools to explore ultra-deep pyrosequencing (UDPS) data, to test these tools, and to use them to determine the optimum error threshold, and to compare results from UDPS and cloning based sequencing (CBS).

Four serum samples, infected with either genotype D or E, from HBeAg-positive and HBeAg-negative patients were randomly selected. UDPS and CBS were used to sequence the basic core promoter/precore region of HBV. Two online bioinformatics tools, the "Deep Threshold Tool" and the "Rosetta Tool" (http://hvdr.bioinf.wits.ac.za/tools/), were built to test and analyze the generated data.

A total of 10952 reads were generated by UDPS on the 454 GS Junior platform. In the four samples, substitutions, detected at 0.5% threshold or above, were identified at 39 unique positions, 25 of which were non-synonymous mutations. Sample #2 (HBeAg-negative, genotype D) had substitutions in 26 positions, followed by sample #1 (HBeAg-negative, genotype E) in 12 positions, sample #3 (HBeAg-positive, genotype D) in 7 positions and sample #4 (HBeAg-positive, genotype E) in only four positions. The ratio of nucleotide substitutions between isolates from HBeAg-negative and HBeAg-positive patients was 3.5 ∶ 1. Compared to genotype E isolates, genotype D isolates showed greater variation in the X, basic core promoter/precore and core regions. Only 18 of the 39 positions identified by UDPS were detected by CBS, which detected 14 of the 25 non-synonymous mutations detected by UDPS.

UDPS data should be approached with caution. Appropriate curation of read data is required prior to analysis, in order to clean the data and eliminate artefacts. CBS detected fewer than 50% of the substitutions detected by UDPS. Furthermore it is important that the appropriate consensus (reference) sequence is used in order to identify variants correctly.