A Study On Tools For Preprocessing The Experimental Results Of Phage Display

Phage display is a high-throughput technique to obtain peptides and proteins withspecific affinities to given targets. This technique has been widely used to not onlyexplore the sites and networks of protein-protein interaction, but also develop newdiagnostics, therapeutics and vaccines. However, target-unrelated peptides (TUPs),either selection-related or propagation-related, often creep into the experimental resultsof phage display due to several intrinsic faults of panning systems and phage libraries.The improvement of experimental methods can not make its results free of TUPs. Thus,computational tools may be an alternative and more convenient way to clean theexperimental results. In this thesis, I finish the following projects:First, I develop a detection tool based on known TUP motifs. The tool is calledTUPScan. It is designed to check whether peptides gained by phage display technologymatch any known TUP motif. If a peptide has a known TUP motif, this peptide is verylikely to be a target-unrelated peptide. Preliminary tests show that TUPScan is efficientand capable of improving the accuracy of mimotope-based epitope mapping. It is alsohelpful for the development of mimotope-based vaccines.Second, two database-based detection tool, i.e. MimoSearch and MimoBlast, aredeveloped to report TUPs without known TUP motifs. The database is called MimoDB,which collects experimental results from phage display technology. Using the MimoDBdatabase as a virtual and comprehensive control in background, the MimoSearch tool iscapable of checking if any submitted peptide has also been gained by other groups withvarious targets. The MimoBlast tool is designed to check if there are peptides in theMimoDB database similar to the peptides user submitted. Completely identical orhighly similar peptides obtained with various targets might also be TUPs.Third, I develop two prediction tools based on support vector machine. Due todeficiency of standard data, the forementioned tools are only reminders rather than realpredictors. With the accumulation of data in the phage display field, two predictorsbased on support vector machine, i.e. PhD7Faster and AvidinBinder are built. ThePhD7Faster tool is capable of predicting if phages from the Ph.D.-7library might grow faster. Thus it is helpful for excluding propagation-related TUPs. The AvidinBinder toolis capable of predicting peptides that can bind to proteins of avidin family, the mostfrequent selection-related TUPs.Finally, all the tools are integrated into the SAROTUP suite, which provides onlineservices for free. Only inputting the sequence information, users can directly get thereport results about TUPs.Above all, several report tools of TUPs are developed and then integrated into auser-friendly web server (SAROTUP, http://immunet.cn/sarotup). It can assistresearchers on new scientific findings. The web programs and database constructed inthis study will become to be important tools and data resources for related researchersaround the world.