def main(df,model_df,T_LibCounts,T_mRNACounts,start=0,end=None): df = evaluate_model.main(df,model_df,left=start,right=None) #We assume only noise is binomial noise(which we approx as poisson) mymodel = Models.PoissonNoise() #calculate new expression levels based on energies of each sequence. libcounts,expcounts = mymodel.gennoisyexp(df,T_LibCounts,T_mRNACounts) return libcounts,expcounts
def main(df, mp, noisetype, npar, nbins, sequence_library=True, start=0, end=None):
# validate noise parameters
if not isinstance(npar, list):
raise SortSeqError("Noise parameters must be given as a list")
if noisetype == "Normal":
if len(npar) != 1:
raise SortSeqError(
"""For a normal noise model, there must be one
input parameter (width of normal distribution)"""
)
if noisetype == "LogNormal":
if len(npar) != 2:
raise SortSeqError(
"""For a LogNormal noise model there must
be 2 input parameters"""
)
if nbins <= 1:
raise SortSeqError("number of bins must be greater than 1")
# generate predicted energy of each sequence.
df = evaluate_model.main(df, mp, left=start, right=None)
# Determine model type to use for noise
if noisetype == "LogNormal":
NoiseModelSort = Models.LogNormalNoise(npar)
elif noisetype == "Normal":
NoiseModelSort = Models.NormalNoise(npar)
elif noisetype == "None":
NoiseModelSort = Models.NormalNoise([1e-16])
else:
NoiseModelSort = Models.CustomModel(noisetype, npar)
# Apply noise to our calculated energies
noisyexp, listnoisyexp = NoiseModelSort.genlist(df)
# Determine Expression Cutoffs for bins
noisyexp.sort()
cutoffs = list(noisyexp[np.linspace(0, len(noisyexp), nbins, endpoint=False, dtype=int)])
cutoffs.append(np.inf)
seqs_arr = np.zeros([len(listnoisyexp), nbins], dtype=int)
# split sequence into bins based on calculated cutoffs
for i, entry in enumerate(listnoisyexp):
seqs_arr[i, :] = np.histogram(entry, bins=cutoffs)[0]
col_labels = ["ct_" + str(i + 1) for i in range(nbins)]
if sequence_library:
df["ct_0"] = utils.sample(df["ct"], int(df["ct"].sum() / nbins))
output_df = pd.concat([df, pd.DataFrame(seqs_arr, columns=col_labels)], axis=1)
col_labels = utils.get_column_headers(output_df)
output_df["ct"] = output_df[col_labels].sum(axis=1)
output_df = output_df.drop("val", axis=1)
return output_df