My lab uses both experimental approaches and bioinformatics to answer questions about viral pathogenesis. We use unbiased molecular approaches to discover novel viruses in clinical materials of interest where there is evidence of acute infection, yet traditional diagnostics do not identify a pathogen. Recently medicine and science have come through an era when it was thought that infectious diseases would soon be largely eradicated or controlled with vaccines and antimicrobials. The AIDS pandemic and the discovery of HIV were a substantial shock to that view. Furthermore, every year new associations are discovered between microbial agents and diseases that were thought to have non-infectious etiologies.

We use two experimental techniques, coupled with a powerful informatics framework that leverages all the viral sequence in GenBank, to directly interrogate interesting specimens that may contain an undetected virus. First, carefully designed long-oligo microarrays can detect patterns that are similar to expected hybridization patterns of know viral taxa. Because of binding energetics on the array, this approach readily detects pathogens which have point mutations that disrupt the binding of validated PCR primers. In addition to detecting such escape-mutants, this approach has been used to discover several novel human and animal viruses over the past few years. The second experimental approach we use is second-generation sequencing of shotgun libraries made directly from clinically important samples. This yields a very large number of relatively short sequence reads which can be directly examined to identify likely viral sequence.

Each of these experimental tactics uses a metagenomic approach, examining RNA and DNA from a complex environment; in our case directly from clinical specimens. The data from both experiments are complex, and sophisticated analysis is required to filter host sequence and recognize enrichment of sequences that may be indicative of the presence of a pathogen. On the gigabase scale, the problem is similar to finding the proverbial needle in a haystack. We use bioinformatics in several powerful ways to find the needle. Every probe on our microarrays is designed to recognize sequence pattern conserved over a particular group of viruses and groups of bright spots can be built up into evidence for the presence of a virus from a particular group. It is critical for analysis of the microarray data or ultra high-throughput sequencing that viral sequence in GenBank and other public databases is coherently linked to a sequence based taxonomy. Development of such a taxonomy is essential and on going in the lab.