Most microbial genome finishing projects at the Joint Genome Institute require the use of a multitude of molecular techniques to achieve a finished genome. The majority of these techniques are applied to sequencing through gaps in the assembly. Traditional shotgun sequencing is known to have difficulty in both cloning of A/T rich regions and sequencing of G/C rich regions. To help alleviate this problem we have applied the 454 sequencing platform as another tool for gap closure. Although 454 Sequencing has been shown to have difficulty with homopolymer stretches of nucleotides, it does not have the same biases as shotgun sequencing. Therefore we feel these two approaches together can be complementary. We have developed a protocol in which gap-spanning fosmids are pooled together, from one or more projects. This DNA is sequenced with 454 and assemblied using the Newbler Assembler and the resulting contigs are added into their respective projects. One 60x60 picotiter chip can yield as much as 32 Mb, allowing for the pooling of 20-28 fosmids with and average read depth of 20-28X. We chose fosmids that spanned gaps in a single microbial genome assembly as well as gap spanning fosmids within a given scaffold of a metagenomic sequencing project. We are also applying 454 technology to whole genome shotgun sequencing to assist with poorly assembled projects from Sanger sequencing alone.

The introduction of pyrosequencing-based sequencing platforms such as 454-sequencing along with the Sanger technology for whole-genome shotgun sequencing has provided an alternative way to produce cost-effective sequence data. For microbial genome finishing, the 454/Sanger hybrid approach has made a considerable impact in reducing the time required to close a genome. However, errors within the 454 sequencing data remain to be addressed to make sure that final consensus represents high quality sequence. JGI is working on further cost reduction of sequencing and finishing processes and is planning significant cut of Sanger sequencing. This will increase the portion of the genome covered with pyrosequence only. Thus the analysis of errors produced by this new technology becomes very important. We conducted a study to assess the quality of the 454 sequence data in order to determine its impact on error rate, and to devise a strategy to correct or reduce the errors. The study analyzed 29 microbial genomes containing both Sanger and 454 only assemblies to look for mismatches in the 454 sequence data. We examined both homopolymer tracts and non-homopolymer regions with respect to read depth and quality assignment. An overview of our findings and strategy for reducing the 454 errors will be presented.