Structure, is a widely used software tool to investigate population genetic structure with multi-locus genotyping data. The software uses an iterative algorithm to group individuals into \"K\" clusters, representing possibly K genetically distinct subpopulations. The serial implementation of this programme is processor-intensive even with small datasets. We describe an implementation of the program within a parallel framework. Speedup was achieved by running different replicates and values of K on each node of the cluster. A web-based user-oriented GUI has been implemented in PHP, through which the user can specify input parameters for the programme. The number of processors to be used can be specified in the background command. A web-based visualization tool \"Visualstruct\", written in PHP (HTML and Java script embedded), allows for the graphical display of population clusters output from Structure, where each individual may be visualized as a line segment with K colors defining its possible genomic composition with respect to the K genetic sub-populations. The advantage over available programs is in the increased number of individuals that can be visualized. The analyses of real datasets indicate a speedup of up to four, when comparing the speed of execution on clusters of eight processors with the speed of execution on one desktop. The software package is freely available to interested users upon request.
Journal of Integrative Bioinformatics (2008) 5 (1): 89 [DOI 10.2390/biecoll-jib-2008-89]
WebStruct and VisualStruct: web interfaces and visualization for Structure software implemented in a cluster environment