parative genomics. Genomic information was not employed in the initial definition of Select Agents. However, from the outset it was asked, to what degree can genomic sequences be used to detect Select Agents now and, for the future, to what degree can genomic sequences be used to predict potentially dangerous sequences in naturally occurring, genetically manipulated or synthetically derived microorganisms? These questions are the focus of this committee.
In examining the sequences from several hundred bacterial genomes, it is impressive just how much diversity is seen—even in different genomes from the same bacterial species. If one considers the general biological criteria used to designate a microorganism as a Select Agent and then asks, to what degree could these attributes be deduced from sequence, the answer is quite clear. Prediction of pathogenicity, transmissibility, ease of preparation, and ease of dissemination is not possible now or in the foreseeable future (see Table 1.1). In part, this is because we lack the basic biological information, and in part it is because our current predictive algorithms are not sufficiently robust.
Infection of a susceptible host by a Select Agent may lead to morbidity and mortality by many different mechanisms, be it “hemorrhagic fever” due to some viral infections or neurological disease following exposure to a bacterial toxin. With the exception of some toxins, the genetic basis for the disease or death that may follow infection with or intoxication by a Select Agent is, however, not well defined. Pathogenicity seen in a susceptible host is the result of a complex interaction between a pathogen and a host defense system, as well as an environmental context (e.g. age, sex, nutrition, health, immune status, and others). As will be discussed subsequently, pathogenicity of an organism may be the result of a specific sequence and gene, or more frequently the result of interactions between several genes, various sequences, structural characteristics, and host characteristics. There are too many variables involved on the host side alone to be able to accurately predict whether any given nucleic acid change in the pathogen will involve greater or lesser pathogenicity. The complexity of these systems argues against a simple gene-sequence basis for “predictive oversight of Select Agents” without substantial new information.
As will be discussed in Chapter 2, there is no current way in which a complex biological factor such as pathogenicity can be predicted from genome sequence. Predicting the function of an individual protein, or single microorganism is daunting. Moreover, the nature of infectious disease is such that accurate prediction of microbial pathogenicity is not possible without information concerning the host and the environmental context.