Novel algorithms for PFGE bacterial typing: Number of co-migrated DNA fragments, linking PFGE to WGS results and computer simulations for evaluation of PulseNet international typing protocols
BackgroundStandard protocols for Pulsed-field gel electrophoresis (PFGE) were adopted and being used in a global scale for surveillance of many bacterial food-borne diseases. Matched PFGE bands are considered regardless of co-migration of different DNA fragments. Molecular epidemiology is turning toward whole genome sequencing (WGS). Although, WGS results can be digested In-silico, PFGE and WGS data are being compared separately. We describe a new image analysis algorithm that enables identification of how many DNA fragments co-migrate during PFGE. We built a database that compare described PFGE results to in-silico obtained digestion models (from WGS). Reliability of the method was assessed in-silico using novel computer simulation approach. From WGS, 1,816 digestion model (DMs) were obtained as recommended by PulseNet international. Simulation codes were designed to predict PFGE profiles when DMs are separated at 5% PFGE resolution in addition to expected co-migration levels. ResultsPFGE simulation has shown that about 35% of DNA fragments co-migrate at 5% PFGE resolution. Similar result was obtained when wet-lab PFGE profiles were analyzed using image analysis algorithm mentioned earlier. When image analysis results were compared to DMs, results returned by geltowgs.uofk.edu database revealed reasonable relatedness to DMs. In terms of number of PFGE typable DNA fragments, 45,517 were typable (representing 46.54% out of 97,801). Previously mentioned typable fragments (in terms of typable sizes) comprised 91.24% of the sum of nucleotides of all chromosomes tested (7.24 billion bp). However, significant variations were shown within and between different digestion protocols. ConclusionsIdentification of co-migration levels will reveal the third dimension of PFGE profiles. This will provide a better way for evaluating isolate relationships. Linking old PFGE results to WGS by means of simulation demonstrated here will provide a chance to link millions of PFGE epidemiological data accumulated during the last 24 years to the new WGS era. Evaluation of population dynamics of pathogenic bacteria will be deeper through space and time. Selection of restriction enzymes for PFGE typing will have a powerful in-silico evaluation tool.