WWU Human MicroBiome Group R Package

Description Details Generating data Modeling data

The HMBGR package provides functions for generating logistic growth time series data, and modelling it using a class of least-squares linear models to recover population parameters.

Originally designed for use by the Human MicroBiome Group at Western Carolina University and Warren Wilson University.

There are several functions provided to generate logistic growth time series data to play with. These functions can generate the data using several different methods, such as an ODE solver, the analytic solution, and the Euler method discretized approximation. Additionally, noise can be added.

Currently, data can be generated to simulate logistic growth of one population, which can be defined either using two parameters, the intrinsic growth rate and the carrying capacity alone, or it can be normalized to a carrying capacity of 1, and specified using the growth rate alone.

In the future, these functions will likely be amalgamated into one function with the option to specify arguments to choose the functionality. For now, the correct function must be selected to do the job required.

Additionally, there are several functions which provide least squares methods to attempt to recover the population parameters from logistic growth time series data. Functions exist for recovering either both the growth rate and the carrying capacity, or for recovering the growth rate alone. Tikhonov regularization, a form of smoothing, is implemented as well.

In our experiment, the parameters can be sucessfully recovered via least squares regression if there is no noise in the data, but even small amounts of noise make smoothing necessary to obtain reasonable estimates. Carrying capacity tends to be difficult to accurately recover.

In the future, these functions will likely be amalgamated into one function with the option to specify arguments to choose the functionality. For now, the correct function must be selected to do the job required.

wz-billings/HMBGR documentation built on May 15, 2020, 5:44 a.m.