Plot all recolonisations in 'plot_daisie_mainland_data'
richelbilderbeek opened this issue · comments
From Issue 62 here I quote:
[...] In this example the ideal data should plot a recolonisation, but only plots one of the species from the all_colonisations.
set.seed(
1,
kind = "Mersenne-Twister",
normal.kind = "Inversion",
sample.kind = "Rejection"
)
daisie_mainland_data <- DAISIEmainland::sim_island_with_mainland(
total_time = 1,
m = 50,
island_pars = c(1.0, 0.5, 10, 0.1, 0.5),
mainland_ex = 2,
mainland_sample_prob = 1,
mainland_sample_type = "complete",
replicates = 1,
verbose = FALSE
)
DAISIEmainland::plot_daisie_mainland_data(
daisie_mainland_data = daisie_mainland_data,
replicate_index = 1
)
So, this is how the plot looks like, after fixing #64:
The first notes:
In code:
> names(daisie_mainland_data$empirical_multi_daisie_data[[1]][[3]])
[1] "branching_times" "stac" "missing_species"
> names(daisie_mainland_data$ideal_multi_daisie_data[[1]][[3]])
[1] "branching_times" "stac" "missing_species"
[4] "all_colonisations"And:
> daisie_mainland_data$empirical_multi_daisie_data[[1]][[3]]
$branching_times
[1] 1.0000000 0.7579685 0.6003790
$stac
[1] 2
$missing_species
[1] 0which should look like:
> daisie_mainland_data$ideal_multi_daisie_data[[1]][[3]]
$branching_times
[1] 1.000000 0.600379
$stac
[1] 3
$missing_species
[1] 0
$all_colonisations
$all_colonisations[[1]]
$all_colonisations[[1]]$event_times
[1] 1.000000 0.600379
$all_colonisations[[1]]$species_type
[1] "A"
$all_colonisations[[2]]
$all_colonisations[[2]]$event_times
[1] 1.0000000 0.1847087
$all_colonisations[[2]]$species_type
[1] "A"which should look like (note that the lines are from the same species, hence the same color):
Now with some notes:
Indeed, the recolonisation does not show up in the result of daisie_data_to_tables:
There are two versions shown, the upper is the one-line-per-re-/-colonist, the lower is the one-line-is-beautiful setup.
@joshwlambert: am I correct that the figure above shows how it should like like? If yes, which version do you prefer?
- The version that is easiest to implement for you
- The upper version
- The lower version
Per email, Josh picked the lower version.
@richelbilderbeek Just to clarify, the lower version is the one line per colonist and not the recolonists plotted on the same line?
@joshwlambert, happy to clarify and be checked :-)
The 'lower version' I meant is like this (I adapted the picture from above) this picture, where a clade would like like this:
What happened is this clade:
- colonisation at t = 0.6
- re-colonisation at t = 0.2
- no branching events
So, does the drawing match expectations?
@joshwlambert well, maybe there is a branching event. Question is, how would you plot this DAISIE data object (it is daisie_mainland_data$ideal_multi_daisie_data[[1]][[3]] in the tests above):
$branching_times
[1] 1.000000 0.600379
$stac
[1] 3
$missing_species
[1] 0
$all_colonisations
$all_colonisations[[1]]
$all_colonisations[[1]]$event_times
[1] 1.000000 0.600379
$all_colonisations[[1]]$species_type
[1] "A"
$all_colonisations[[2]]
$all_colonisations[[2]]$event_times
[1] 1.0000000 0.1847087
$all_colonisations[[2]]$species_type
[1] "A"@richelbilderbeek I would plot the two colonisations in the all_colonistations as two separate lines. The colour of which would depend on their endemicity given by the species_type in all_colonisations.
@joshwlambert so this would be a good sketch?
OK, zooming in on the problem, I added this test to test-plot_daisie_data.R:
test_that("Issue #68: plot all recolonisations", {
set.seed(
1,
kind = "Mersenne-Twister",
normal.kind = "Inversion",
sample.kind = "Rejection"
)
daisie_mainland_data <- DAISIEmainland::sim_island_with_mainland(
total_time = 1,
m = 50,
island_pars = c(1.0, 0.5, 10, 0.1, 0.5),
mainland_ex = 2,
mainland_sample_prob = 1,
mainland_sample_type = "complete",
replicates = 1,
verbose = FALSE
)
daisie_data <- daisie_mainland_data$ideal_multi_daisie_data[[1]]
plot_daisie_data(daisie_data)
})This is the problem:
Using vertical lines solves the problem in a way:
@richelbilderbeek I have accepted the PR. Thanks for the nice plots. However, I would like the recolonists plotted as separate lines, each with their own colonisation time and not linked by any branching events. Would this be possible.
@joshwlambert I think yes, as Saturday I seem to have plenty of time to fix that, as well as fixing the vertical branches being misplaced!
This will be harder: the algorithm to spread out branches is one thing, doing the same for colonists, that may branch out as well, may be hard! Let's first write a solid test 👼
This is an interesting setting:
test_that("Issue #68: plot all recolonisations with many branches", {
set.seed(
631,
kind = "Mersenne-Twister",
normal.kind = "Inversion",
sample.kind = "Rejection"
)
daisie_mainland_data <- DAISIEmainland::sim_island_with_mainland(
total_time = 1,
m = 10,
island_pars = c(1, 1, 10, 0.1, 1),
mainland_ex = 1,
mainland_sample_prob = 1,
mainland_sample_type = "complete",
replicates = 1,
verbose = FALSE
)
DAISIEmainland::plot_daisie_mainland_data(
daisie_mainland_data = daisie_mainland_data,
replicate_index = 1
)
# Plots nicely
daisie_data <- daisie_mainland_data$ideal_multi_daisie_data[[1]]
plot_daisie_data(daisie_data)
# Plots nicely when there are no colonisations
daisie_data <- daisie_mainland_data$empirical_multi_daisie_data[[1]]
plot_daisie_data(daisie_data)
})
Data:
> daisie_data
[[1]]
[[1]]$island_age
[1] 1
[[1]]$not_present
[1] 9
[[2]]
[[2]]$branching_times
[1] 1.0000000 0.9115751 0.5076183 0.3368938
[[2]]$stac
[1] 3
[[2]]$missing_species
[1] 0
[[2]]$all_colonisations
[[2]]$all_colonisations[[1]]
[[2]]$all_colonisations[[1]]$event_times
[1] 1.0000000 0.9115751 0.5076183 0.3368938
[[2]]$all_colonisations[[1]]$species_type
[1] "C"
[[2]]$all_colonisations[[2]]
[[2]]$all_colonisations[[2]]$event_times
[1] 1.000000 0.344173
[[2]]$all_colonisations[[2]]$species_type
[1] "A"
OK, that one should display as:
Here is the best one yet:
test_that("Multiple recolonisations", { # nolint indeed, this is complex :-)
skip("Only run locally")
seed <- 1912
set.seed(
seed,
kind = "Mersenne-Twister",
normal.kind = "Inversion",
sample.kind = "Rejection"
)
daisie_mainland_data <- sim_island_with_mainland(
total_time = 1,
m = 10,
island_pars = c(1, 0.1, 30.0, 1.0, 5.0),
mainland_ex = 1,
mainland_sample_prob = 1,
mainland_sample_type = "complete",
replicates = 1,
verbose = FALSE
)
ideal_daisie_data <- daisie_mainland_data$ideal_multi_daisie_data[[1]]
clade_index <- 9
n_colonisations <- length(ideal_daisie_data[[clade_index]]$all_colonisations) # nolint indeed a long line
n_branches <- length(ideal_daisie_data[[clade_index]]$branching_times) - 1 # nolint indeed a long line
expect_true(n_colonisations >= 3 && n_branches >= n_colonisations * 2)
interesting_clade <- ideal_daisie_data[clade_index]
plot_daisie_data(ideal_daisie_data)
simplified_ideal_daisie_data <- list()
simplified_ideal_daisie_data[[1]] <- ideal_daisie_data[[1]]
simplified_ideal_daisie_data[[2]] <- ideal_daisie_data[[9]]
plot_daisie_data(daisie_data = simplified_ideal_daisie_data)
})
with
> simplified_ideal_daisie_data
[[1]]
[[1]]$island_age
[1] 1
[[1]]$not_present
[1] 5
[[2]]
[[2]]$branching_times
[1] 1.00000000 0.81360371 0.59927594 0.54433504 0.43080142 0.11137935 0.04632135
[[2]]$stac
[1] 3
[[2]]$missing_species
[1] 0
[[2]]$all_colonisations
[[2]]$all_colonisations[[1]]
[[2]]$all_colonisations[[1]]$event_times
[1] 1.0000000 0.8136037 0.5992759 0.4308014
[[2]]$all_colonisations[[1]]$species_type
[1] "C"
[[2]]$all_colonisations[[2]]
[[2]]$all_colonisations[[2]]$event_times
[1] 1.00000000 0.54433504 0.04632135
[[2]]$all_colonisations[[2]]$species_type
[1] "C"
[[2]]$all_colonisations[[3]]
[[2]]$all_colonisations[[3]]$event_times
[1] 1.0000000 0.3858606 0.1113793
[[2]]$all_colonisations[[3]]$species_type
[1] "C"
I guess this is how it should look?
Getting there:
Done! It is nearly beautiful:
