BA3-SNPs: Contemporary migration reconfigured in BayesAss for next-generation sequence data

Quantifying demographic independence is a vital step in establishing potential conservation units for a species in that it effectively distinguishes migration from within-population reproduction. This is an important aspect because it allows for an accurate estimate of recruitment. For example, populations may be designated as 'management units' (=MUs) if indeed population growth results from local demography rather than immigration. Of additional interest is the calculation of immigrant ancestry and ascertainment of the temporal context over which immigration occurred. This is because MUs depend largely upon local (self-sustaining) birth and death rates, and the quantification of ancestry is necessary to validate demographic independence. Dispersal rate is also of immediate interest to conservation biologists, and can be assessed by quantifying genetic divergence among populations. The capacity with which to gauge these benchmarks has now been extended herein to genome-wide molecular data, in an attempt to adjust an analytical tool that was until now intractable for the next generation sequencing data. In this study, a popular legacy program for migrant detection (i.e. BayesAss3) has been modified to accept SNP (single nucleotide polymorphism) data. We validated BA3-SNPs using empirical data to demonstrate its suitability for both high-performance and desktop computing environments. We also facilitate high analytical throughput by presenting a binary search algorithm that automates MCMC (Markov chain Monte Carlo) parameter tuning. Our BA3-SNPs-autotune program required five or fewer rounds of optimization for 99% of input files, with acceptable mixing parameters derived in 100% of our test cases. Runtime for BA3-SNPs is a function of the number of loci analysed. Benchmarking yielded an average runtime <32 hr (10 million MCMC generations) for datasets containing thousands of SNPs. The BA3 algorithm remains a viable option for analysing modern SNP datasets. Source code (C++ and Python) is released publicly under the GNU General Public License v3.0, and is available for download (Linux and Mac OSX) from the following URL: .