What is BaySICS?


BaySICS is a program aimed to perform statistical inference by approximate Bayesian computation (ABC), employing coalescent simulations as the framework model and a DNA alignment as the observed data (as well as other information).

The goal is to obtain parameters estimations or hypothesis contrasts. The products are point and interval estimations of parameters of interest, or an acceptance-rejection decision for competing hypothesis or models that take the form of simulated scenarios. A numerical indicator of the confidence of a model/scenario choice is also part of the pursued products of the analysis.

The simulated scenarios are usually sophisticated and there are many sources of noise, uncertainty or bias that can affect the analysis in unpredictable ways. For that reason, some grants are usually required in addition to reasonableness of the assumptions in order to have a measure of confidence in the procedure. BaySICS incorporates an algorithm that test the statistical properties of the analysis taking datasets obtained from simulations as if they were the real data. Those pseudo observed datasets (PODs) provide important information to support the conclusions of the main analyses. 

Why using BaySICS?

  • BaySICS is user-friendly. BaySICS provides a graphic interface (GUI) with all the options and analysis reachable from the GUI.
  • BaySICS is parallelizable. The coalescent simulator can run as many groups of simulations as desired, and automatically puts on queue those exceeding the number of cores in the system. The analysis programs can analyze the parallel reference tables without additional work.
  • BaySICS is powerful. BaySICS offers a number of advantages in every stage of an ABC procedure, from simulations to validation analyses, including flexible simulations designs, an MCMC algorithm, and validation analysis by means of pseudo observed datasets (PODs).

News in Evolutionary Research


Subpages (1): Other programs