def fates(self, populations):
"""
Computes fates for each population
Parameters
----------
self : wot.TransportMapModel
The TransportMapModel used to find fates
populations : list of wot.Population
The target populations such as ones from self.population_from_cell_sets. The populations must be from the same time.
Returns
-------
fates : anndata.AnnData
Rows : all cells, Columns : populations index. At point (i, j) : the probability that cell i belongs to population j
"""
start_day = wot.tmap.unique_timepoint(
*populations) # check for unique timepoint
populations = Population.copy(*populations,
normalize=False,
add_missing=True)
pop_names = [pop.name for pop in populations]
results = []
results.insert(0, np.array([pop.p for pop in populations]).T)
while self.can_pull_back(*populations):
populations = self.pull_back(*populations,
as_list=True,
normalize=False)
results.insert(0, np.array([pop.p for pop in populations]).T)
X = np.concatenate(results)
X /= X.sum(axis=1, keepdims=1)
obs = self.meta.copy()
obs = obs[obs['day'] <= start_day]
return anndata.AnnData(X=X, obs=obs, var=pd.DataFrame(index=pop_names))
def get_population(ids_el):
cell_indices = df.index.get_indexer_for(ids_el)
cell_indices = cell_indices[cell_indices > -1]
if len(cell_indices) is 0:
return None
p = np.zeros(len(df), dtype=np.float64)
p[cell_indices] = 1.0
return Population(day, p / np.sum(p))
def transition_table(self, start_populations, end_populations):
"""
Computes a transition table from the starting populations to the ending populations
Parameters
----------
self : wot.TransportMapModel
The TransportMapModel
start_populations : list of wot.Population
The target populations such as ones from self.population_from_cell_sets. THe populations must be from the same time.
Returns
-------
transition table : anndata.AnnData
Rows : starting populations, Columns : ending populations.
"""
# add "other" population if any cells are missing across all populations
start_time = wot.tmap.unique_timepoint(*start_populations)
start_populations = Population.copy(*start_populations,
normalize=False,
add_missing=True)
end_populations = Population.copy(*end_populations,
normalize=False,
add_missing=True)
wot.tmap.unique_timepoint(
*end_populations) # check for unique timepoint
populations = end_populations
results = []
results.insert(0, np.array([pop.p for pop in populations]).T)
while self.can_pull_back(*populations) and wot.tmap.unique_timepoint(
*populations) > start_time:
populations = self.pull_back(*populations,
as_list=True,
normalize=False)
end_p = np.vstack([pop.p for pop in populations])
start_p = np.vstack([pop.p for pop in start_populations])
p = (start_p @ end_p.T)
p = p / p.sum()
return anndata.AnnData(
X=p,
obs=pd.DataFrame(index=[p.name for p in start_populations]),
var=pd.DataFrame(index=[p.name for p in end_populations]))
def trajectories(self, populations):
"""
Computes a trajectory for each population
Parameters
----------
self : wot.TransportMapModel
The TransportMapModel used to find ancestors and descendants of the population
populations : list of wot.Population
The target populations such as ones from self.population_from_cell_sets. THe populations must be from the same time.
Returns
-------
trajectories : anndata.AnnData
Rows : all cells, Columns : populations index. At point (i, j) : the probability that cell i is an
ancestor/descendant of population j
"""
wot.tmap.unique_timepoint(*populations) # check for unique timepoint
trajectories = []
populations = Population.copy(*populations,
normalize=True,
add_missing=False)
population_names = [p.name for p in populations]
initial_populations = populations
def update(head, populations_to_update):
idx = 0 if head else len(trajectories)
trajectories.insert(
idx,
np.array([pop.p for pop in populations_to_update]).T)
update(True, populations)
while self.can_pull_back(*populations):
populations = self.pull_back(*populations, as_list=True)
update(True, populations)
populations = initial_populations
while self.can_push_forward(*populations):
populations = self.push_forward(*populations, as_list=True)
update(False, populations)
return anndata.AnnData(X=np.concatenate(trajectories),
obs=self.meta.copy(),
var=pd.DataFrame(index=population_names))
def pull_back(self,
*populations,
to_time=None,
normalize=True,
as_list=False):
"""
Pulls the population back through the computed transport maps
Parameters
----------
*populations : wot.Population
Measure over the cells at a given timepoint to be pushed forward.
to_time : int or float, optional
Destination timepoint to pull back to.
normalize : bool, optional, default: True
Wether to normalize to a probability distribution or keep growth.
as_list : bool, optional, default: False
Wether to return a listof length 1 when a single element is passed, or a Population
Returns
-------
result : wot.Population
The pull back of the input population through the proper transport map.
Array of populations if several populations were given as input.
Raises
------
ValueError
If there is no previous timepoint to pull the population back.
ValueError
If several populations are given as input but dot live in the same timepoint.
Examples
--------
>>> self.pull_back(pop, to_time = 0) # -> wot.Population
Pushing several populations at once
>>> self.pull_back(pop1, pop2, pop3) # -> list of wot.Population
Pulling back after pushing forward
>>> self.pull_back(self.push_forward(pop))
Same, but several populations at once
>>> self.pull_back(* self.push_forward(pop1, pop2, pop3))
"""
i = self.timepoints.index(wot.tmap.unique_timepoint(*populations))
j = i - 1 if to_time is None else self.timepoints.index(to_time)
if i == -1:
raise ValueError("Timepoint not found")
if i == 0:
raise ValueError("No previous timepoints. Unable to pull back")
if j == -1:
raise ValueError("Destination timepoint not found")
if i < j:
raise ValueError(
"Destination timepoint is after source. Unable to pull back")
p = np.vstack([pop.p for pop in populations])
while i > j:
t1 = self.timepoints[i]
t0 = self.timepoints[i - 1]
tmap = self.get_coupling(t0, t1)
p = (tmap.X @ p.T).T
if normalize:
p = (p.T / np.sum(p, axis=1)).T
i -= 1
result = [
Population(self.timepoints[i], p[k, :]) for k in range(p.shape[0])
]
if len(result) == 1 and not as_list:
return result[0]
else:
return result
