CBB-Seminar, Tues. June 15, 11:00 am, 8th floor conference room:

Paul Kitts

NEW TOOLS TO MINIMIZE THE INCIDENCE AND IMPACT OF VECTOR CONTAMINATION

Maintaing high sequence quality in public databases is critical 
given the increasing role of bioinformatics in the scientific arena. 
Database sequences that contain segments of foreign origin can 
confound the interpretation of any analysis performed using a data 
set that includes such contaminated sequences. For example, a 
similarity search can produce hits that are based only on shared 
segments of foreign sequence. Although nucleotide sequence analyses 
are most susceptible to the effects of contamination, protein analyses 
can also be affected since foreign sequence segments are occasionally 
found within a coding region. The most common source of contamination 
is vectors used to clone and sequence the source DNA/RNA. Improved 
methods for detecting sequences derived from vectors will therefore 
help to reduce the incidence and impact of contaminated sequences in 
public databases. To this end, a specialized database of non-redundant 
vector sequences has been developed for use in screening nucleotide 
sequences for contamination. In addition, the similarity search 
parameters for detection of vector contamination have been optimized. 
Finally, criteria for automatic catagorization of segments of potenial 
vector contamination have been defined. The result of this work is a 
quick, sensitive, and descriminating screen for vector contamination. 
The new screening system is currently being used to identify nucleic 
acid sequences with vector contamination before they are added to 
GenBank. Another application for the new screen would be to suppress 
the effects of vector contamination already present in public 
databases by masking vector sequences prior to a similarity search.