Skip to content

Commit

Permalink
minor bug fixes
Browse files Browse the repository at this point in the history
  • Loading branch information
@sjohri20
sjohri20 committed Apr 19, 2023
1 parent 961b948 commit 3ff716c
Showing 1 changed file with 23 additions and 23 deletions.
46 changes: 23 additions & 23 deletions src/beanie/beanie.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -72,7 +72,7 @@ def __init__(self, counts_path: str, metad_path: str, sig_path:str, normalised:b
self.heatmap figure for HeatmapDriverGenes function
self.upsetplot_driver_genes figure for UpsetPlotDriverGenes function
self.upsetplot_signature_genes figure for UpsetPlotSignatureGenes function
self.de_obj DifferentialExpression object for max/custom subsample size
self.de_obj differentialExpression object for max/custom subsample size
self.de_summary dataframes containing the output of DifferentialExpression
self.de_obj_simulation list of DifferentialExpression objects for max/custom subsample siz
self.de_summary_simulation dictionary mapping the subsample size to dataframes generated from DifferentialExpression object
Expand All @@ -81,7 +81,7 @@ def __init__(self, counts_path: str, metad_path: str, sig_path:str, normalised:b
self.d1_all dictionary mapping patients to cell_ids in treatment group A
self.d2_all dictionary mapping patients to cell_ids in treatment group B
self.max_subsample_size
self.group_id_names list of treatment groups names in self.metad
self.group_id_names list of treatment groups names in self.metad
self.top_signatures top 5 most significant and robust genes
self.num_driver_genes number of driver genes for which plots to be made
self.t1_cells
Expand Down Expand Up @@ -291,22 +291,24 @@ def _writeSignatures(self):
return


def SignatureScoring(self, scoring_method="beanie", no_random_sigs=1000, aucell_quantile=0.05):
def SignatureScoring(self, scoring_method="beanie", no_random_sigs=1000):
"""
Function to do signature scoring using in-built scoring functions.
Parameters:
scoring_method choice between beanie (default), mean and combined-z to score the cells.
no_random_sigs the number of random signatures that should be generated for FDR correction
aucell_quantile parameter to indicate the quantile of genes to consider for ROC, if beanie method of scoring is being used.
scoring_method 'beanie' (AUCell-inspired, default), 'mean' (weighted mean) and 'combined-z' (z-score).
no_random_sigs The number of background signatures that should be generated for p-value correction.
"""

# Parameter to indicate the quantile of genes to consider for ROC
aucell_quantile=0.05

self._scoring_method = scoring_method

logging.info("Scoring signatures...")

# Score background signatures
# Score background gene signatures
sorted_genes = pd.Series.sort_values(self.normalised_counts.sum(axis=1))
null_dist_sigs = GenerateNullDistributionSignatures(self.signatures, sorted_genes, self._bins, self.output_dir, no_random_sigs)
self._null_dist_scores = dict()
Expand Down Expand Up @@ -351,18 +353,19 @@ def SignatureScoring(self, scoring_method="beanie", no_random_sigs=1000, aucell_
return


def DifferentialExpression(self, cells_to_subsample_1=None, cells_to_subsample_2=None, alpha=0.05, min_ratio=0.9, subsamples=501, test_name="mwu-test", group_direction = None, **kwargs):
def DifferentialExpression(self, cells_to_subsample_1=None, cells_to_subsample_2=None, alpha=0.05, min_ratio=0.9, subsamples=500, test_name="mwu-test", group_direction = None, **kwargs):
"""
Function for finding out differentially expressed robust and statistically significant signatures.
Parameters:
cells_to_subsample cells that should be subsampled per patient; if no input provided, function to choose the max possible subsample size
cells_to_subsample1 Cells subsampled per sample in group1; by default choose the max possible subsample size.
cells_to_subsample2 Cells subsampled per sample in group2; by default choose the max possible subsample size.
alpha p-value cutoff
min_ratio value of fold_rejection_ratio below which the signature is considered to be non-robust
subsamples number of repeated subsamples in every fold
minimum_expressing_samples minimum number of samples that express gene to be considered
minimum_frac_per_sample minimum fraction of cells expressing for a gene to be considered expressed in a sample
minimum_expression minimum expression value for a gene to be considered expressed in a cell
min_ratio Value of Fold Rejection Ratio (FRR( below which the signature is considered to be non-robust.
subsamples Number of repeated subsamples in every fold. Default = 500.
minimum_expressing_samples Minimum number of samples that express gene signature.
minimum_frac_per_sample Minimum fraction of cells expressing for a gene signature to be considered expressed in a sample.
minimum_expression Minimum expression value for a gene signature to be considered expressed in a cell.
"""
if self._differential_expression_run == True:
Expand Down Expand Up @@ -444,19 +447,16 @@ def DifferentialExpression(self, cells_to_subsample_1=None, cells_to_subsample_2

def GetDifferentialExpressionSummary(self):
if self._differential_expression_run==True:
if self._sig_score_path==None:
return self.de_summary[["log2fold","p","corr_p","corrected_p_inbuilt","nonrobust","direction"]]
else:
return self.de_summary[["log2fold","p","corr_p","nonrobust","direction"]]
return self.de_summary[["log2fold","p","corr_p","nonrobust","direction"]]
else:
raise RuntimeError("Run DifferentialExpression() first.")

def RankGenes(self, group_direction=None):
if self._driver_genes_run==True:
print("DriverGenes() has already been run.")
print("RankGenes() has already been run.")
return

logging.info("Finding Driver Genes...")
logging.info("Ranking Genes...")

if self._differential_expression_run==False:
raise RuntimeError("Run DifferentialExpression() first.")
Expand Down Expand Up @@ -485,7 +485,7 @@ def RankGenes(self, group_direction=None):
def GetRankGenesSummary(self):

if self._driver_genes_run==False:
raise RuntimeError("Run DriverGenes() method first.")
raise RuntimeError("Run RankGenes() method first.")

elif self._differential_expression_run==False:
raise RuntimeError("Run DifferentialExpression() first.")
Expand Down Expand Up @@ -705,7 +705,7 @@ def GeneRankHeatmap(self, signature_names=None, num_genes = 10, **kwargs):
signature_names = self.top_signatures

if self._driver_genes_run==False:
raise RuntimeError("Run DriverGenes() first.")
raise RuntimeError("Run RankGenes() first.")

self.num_driver_genes = num_genes
self.heatmap = dg.GenerateHeatmap(self.normalised_counts.T, self.t1_ids, self.t2_ids, self.d1_all, self.d2_all, self.driver_genes, signature_names, num_genes, **kwargs)
Expand Down Expand Up @@ -733,7 +733,7 @@ def GeneRankUpsetPlot(self, fig_width=None, signature_names=None):
print("Too many signature names to show upset plot")

if self._driver_genes_run==False:
raise RuntimeEror("Run DriverGenes() first.")
raise RuntimeEror("Run RankGenes() first.")

upset_df_prep = pd.DataFrame(columns=self.driver_genes.keys())
for x in self.driver_genes.keys():
Expand Down

0 comments on commit 3ff716c

Please sign in to comment.