def tidy_sort(df):
df = compute_cell_states(df)
df_tidy = df.melt(id_vars=["time", "Infection"], var_name="celltype")
cells = filter_cells(df_tidy, ["Th1_eff", "Tfh_all", "Tr1_all", "nonTfh",
"Total_CD4"])
return df_tidy, cells
def transform(sim, data, timepoints=[9, 30, 60]):
"""
objective function computes difference between data and simulation
Parameters
----------
sim : TYPE
DESCRIPTION.
data : TYPE
DESCRIPTION.
timepoints : TYPE, optional
DESCRIPTION. The default is [9, 30, 60].
Returns
-------
resid : arr
array of data-simulation differences (residuals)
"""
sim = pd.DataFrame(sim, columns=sim.colnames)
# get simulation data at these time points
sim = sim[sim.time.isin(timepoints)]
sim = compute_cell_states(sim)
# only focus on Tfh and non Tfh
sim = sim[["Tfh_all", "nonTfh"]]
# convert to same format as data
sim = sim.melt()
resid = (data.value.values - sim.value.values) / data.eps.values
return resid
def tidy(sim, name):
"""
convert simulation output to long format for plotting
"""
sim = pd.DataFrame(sim, columns=sim.colnames)
# get simulation data at these time points
sim = compute_cell_states(sim)
# only focus on Tfh and non Tfh
sim = sim[["time", "Tfh_all", "nonTfh"]]
sim = sim.melt(id_vars=["time"])
sim["Infection"] = name
return sim