Multi-core computer systems increase performance without increasing the cost exponentially. The main hindrance in the way of utilizing this gain in performance is that the sequential programs are only capable of using just one core at a time. A completely new set of obstacles have to be surmounted, in designing a software capable of running multiple instances of itself or its parts simultaneously thus utilizing the performance benefit of these multi-core computers.The project involved studying the various approaches of parallelization and analyzing the working of already existing simulation software Plabsoft and Plabsim followed by designing of the new parallelized simulation software PopSim. It was mandatory that the new software (i) should not be dependent on any third party libraries; (ii) must be completely backwards compatible with Plabsoft; (iii) the PopSim should be able to execute faster than Plabsoft while giving comparable output; (iv) it should be capable of executing on both Linux and Windows systems; (v) the new software should compile cleanly; (vi) it must be able to compile on compilers not supporting parallelization. New algorithms to handle population genetics problems on multi-core computers were required to be designed and implemented.The PopSim software that has been developed is an add-on package for the "R Statistical Software" coded in C and R. It is a flexible and powerful parallelized simulation software for population genetics and data analysis capable of fully utilizing the multi-core architecture of computers. It is able to handle a broad range of problems concerning plant breeding and genetics. PopSim is able to simulate the various mating scenarios like random mating, selfing, partial selfing, planned crosses, double haploids, top-crosses, single-seed descent and factorials. It provides the means to simulate selections according to selection indices based on molecular marker scores and/or phenotypic values. A range of data analysis routines are provided in PopSim to analyze simulated and experimental datasets for allele and genotype frequencies, genotypic and phenotypic values and variances, molecular genetic diversity, linkage disequilibrium and parameters to optimize marker-assisted backcrossing programs using simulated and experimental datasets.On testing, it was observed that PopSim produced results, which were identical to the results produced by Plabsoft for all the examples. Significant improvement in the speed of execution was observed for examples that utilized the parallelized part of the software. The maximum improvement in speed that was observed was 60% i.e. PopSim only took 40% of the time that Plabsoft took. The load time of PopSim was only 1/3 of the Plabsoft´s load time. In addition, it was also established that PopSim compiles cleanly under both Linux and Windows operating systems without giving any errors or warnings.Finally it could be concluded that PopSim demonstrates a considerable and significant improvement in performance over Plabsoft. It fulfills the requirements and goals set for the project to a large extent.
Verknüpfung zu Publikationen oder weiteren Datensätzen
