post_prob_states: Calculate posterior probability of each states/geographic...
In BioGeoBEARS: BioGeography with Bayesian (and Likelihood) Evolutionary Analysis in R Scripts

Description Usage Arguments Details Value Note Author(s) References See Also Examples

View source: R/BioGeoBEARS_detection_v1.R

This function calculates P(range|data,detection model), i.e. the probability of each possible range, given a prior probability of each range, and the likelihood of each range.

1 2	post_prob_states(prob_of_each_range, condlikes_of_data_on_each_range)

`prob_of_each_range`	The probability of each range, given the prior probability of presence in each area.
`condlikes_of_data_on_each_range`	The probability of the data, conditional on each range (i.e., the likelihood), as found in e.g. a row of the output from `tiplikes_wDetectionModel`.

The prior probability of each range should be considered by the user. Note that putting the same prior on the probability of occurrence in each individual range does NOT mean a flat prior on each state/geographic range. This fact is demonstrated in the function prob_of_states_from_prior_prob_areas.

posterior_probs The posterior probability of each range.

Go BEARS!

Nicholas J. Matzke matzke@berkeley.edu

http://phylo.wikidot.com/matzke-2013-international-biogeography-society-poster http://en.wikipedia.org/wiki/Log_probability

Matzke_2012_IBS

Bottjer_Jablonski_1988

prob_of_states_from_prior_prob_areas, tiplikes_wDetectionModel, rcpp_areas_list_to_states_list, Pdata_given_rangerow, calc_obs_like, mapply, read_detections, read_controls

testval=1

# soft-coded input files
extdata_dir = np(system.file("extdata", package="BioGeoBEARS"))
detects_fn = np(paste(extdata_dir, "/Psychotria_detections_v1.txt", sep=""))
controls_fn = np(paste(extdata_dir, "/Psychotria_controls_v1.txt", sep=""))

detects_df = read_detections(detects_fn, OTUnames=NULL, areanames=NULL, tmpskip=0)
controls_df = read_controls(controls_fn, OTUnames=NULL, areanames=NULL, tmpskip=0)


# Setup
prior_prob_presence = 0.01
areas = c("K", "O", "M", "H")
numareas = length(areas)
maxareas = length(areas)
states_list_0based_index =
rcpp_areas_list_to_states_list(areas=areas, maxareas=maxareas,
                               include_null_range=TRUE)
states_list_0based_index

mean_frequency=0.1
dp=1
fdp=0

tip_condlikes_of_data_on_each_state =
tiplikes_wDetectionModel(states_list_0based_index, numareas=numareas,
detects_df, controls_df, mean_frequency=mean_frequency, dp=dp, fdp=fdp,
null_range_gets_0_like=TRUE)

tip_condlikes_of_data_on_each_state



# To get denominator, just iterate over all the states
# Prior probability
prob_of_each_range = prob_of_states_from_prior_prob_areas(states_list_0based_index,
numareas=numareas,
prior_prob_presence=prior_prob_presence, null_range_gets_0_prob=TRUE,
normalize_probs=TRUE)

# Likelihoods of the data on each range
condlikes_of_data_on_each_range = tip_condlikes_of_data_on_each_state[1,]

posterior_probs = post_prob_states(prob_of_each_range,
                  condlikes_of_data_on_each_range)
posterior_probs

# Should sum to 1
sum(posterior_probs)