Skip to contents

1. Introduction

a1_introduction.Rmd
library(biPOD)
require(dplyr)
#> Loading required package: dplyr
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

Input format

The input for biPOD is a tibble with two columns:

  • time denotes the instant of time in which each observation is collected
  • count contains the values of the observations which should be a proxy of a population size (e.g. number of cells, volume, …)

biPOD comes with a template dataset.

# Load template data
data("xenografts", package = "biPOD")

Initialisation of a new dataset

To use biPOD you need to initialize a bipod object with the initialisation function init.

When you create a new object it is required to explicit the sample name.

Additionally, since biPOD might deal with longitudinal observations coming from different time windows (e.g Natural evolution, Treatment, Relapse,…) divided by a set of breakpoints, the user should also input them if known. (see below)

Example biPOD object

Let’s use one of the sample coming from the xenograft data along with a made up set of breakpoints.

mouse_id <- 529
x <- biPOD::init(
  counts = xenografts %>% dplyr::filter(mouse == mouse_id) %>% dplyr::mutate(count = tumour_volume),
  sample = mouse_id,
  break_points = c(0, 20)
)
#> 
#> ── biPOD - bayesian inference for Population Dynamics ──────────────────────────
#> ℹ Using sample named: 529.
#> ! No group column present in input dataframe! A column will be added.

If you look at the modified counts you will see that the observations have been grouped in different time windows

biPOD::plot_input(x, add_highlights = T)