input_dir) #compare_from=inputfile
#Keep only selected markers
if filter_markers:
selected_markers = read_marker_csv(input_dir)
print(compare_to.columns)
file_cols = compare_to.columns
[x for x in file_cols if x[0].isdigit()]
for x in file_cols:
if x[0].isdigit():
if x not in selected_markers:
compare_to = compare_to.drop(x, axis=1)
print(compare_to.columns)
# compare_to = downsample_data(compare_to, info_run, output_dir) #Customtest_1
compare_to_arc, marker_list = arcsinh_transf(cofactor,
compare_to) #Leeave as default
# compare_to_arc = compare_to #Customtest_2
# marker_list = [x for x in compare_to_arc.columns if x[0].isdigit()] #Customtest_2
print('Sample files:')
print('\n'.join([f for f in input_files]))
print(f'\nReference:\n{denominator}')
print('\nMarkers:')
print('\n'.join([m for m in marker_list]))
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Perform EMD~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~#
#Calculate EMD and save to the output folder (with denominator info run):
# calculate emd and sign the emd score by the difference of median between compare_from and compare_to
emd_df = pd.DataFrame()
emd_infodict = {}
data = fcsparser.parse(file_path, meta_data_only=False)[1]
reg_pnn = re.compile("(\d+Di$)") #Detect if, despite flag
pnn_extracted = [] #columns match PnN pattern
for n in data.columns.values.tolist():
if reg_pnn.search(n):
pnn_extracted.append(n)
if len(pnn_extracted) != 0:
raise fcsparser.api.ParserFeatureNotImplementedError
except fcsparser.api.ParserFeatureNotImplementedError:
print("WARNING: Non-standard .fcs file detected: ", f)
#use rpy2 to read the files and load into python
data = read_rFCS(file_path)[0]
if filter_markers: #Load .csv with the markers to use -> Often PTMs
selected_markers = read_marker_csv(input_dir)
data = data.loc[:, selected_markers] # Remove unwanted markers
data_arc, markers = arcsinh_transf(cofactor, data)
# generate the list of marker-marker pairs for dremi calculation
marker_pairs = [comb for comb in list(permutations(markers, 2))]
for marker_x, marker_y in marker_pairs:
df_info_dict = {}
df_info_dict["file_origin"] = filename
df_info_dict["marker_x"] = marker_x
df_info_dict["marker_y"] = marker_y
df_info_dict["marker_x_marker_y"] = marker_x + '_' + marker_y
df_info_dict["num_of_cells"] = data.shape[0]
if plot == True:
if os.path.isdir(
f'{output_dir}/{info_run}/plots/x={marker_x}-y={marker_y}'
) == False:
###~Co-factor~###
#Check if user wants to filter the markers based on a .csv marker file
cofactor = 5
user_cofactor = yes_or_NO(
"Using alpha=5 for the transformation. Would you like to change this value?"
)
if user_cofactor:
cofactor = int(input("Enter the new alpha to use (5=default): "))
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~#
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Perform transformation~~~~~~~~~~~~~~~~~~~~~~~~~~#
#Identify marker columns
for input_file in filelist:
markers = []
dataset = pd.read_csv(f"{input_dir}/{input_file}", sep='\t')
print("Data read!")
markers.append([x for x in dataset.columns if x[0].isdigit()])
print("Processed columns")
for i in markers:
print("Columns identified as markers: \n", i)
print("Start transform (might take around 10min with larger datasets)")
normalised_dataset = arcsinh_transf(cofactor, dataset)[0]
print("Finished transform")
print(
"Start writing results to file (might take some time with larger datasets)"
)
normalised_dataset.to_csv(f"{output_dir}/arcsinhTRANSF_{input_file}",
index=False,
sep='\t')
downs_inputs = yes_or_NO(
"Multiple input files detected. Would you like to donwsample the number of cells?",
default="YES")
if downs_inputs:
print("Downsampling taking place. Check output folder for more info")
print(no_arc["file_origin"].value_counts())
no_arc = downsample_data(no_arc, info_run, output_dir)
print(no_arc["file_origin"].value_counts())
else:
print("Multiple input files; no downsampling")
else:
print("Only one input file detected; no downsampling")
#Transformation#
#Literature recommends cofactor of 5 for cytof data
arc, cols = arcsinh_transf(cofactor, no_arc)
#Storing marker columns for later use below
#~~~~~~~~~~~~~~~Define the markers used for UMAP calculation~~~~~~~~~~~~~~~~~~#
#Group columns of the dataframe based on the type of measurement
not_markers_cols = [column for column in arc.columns if column not in cols]
all_markers_cols = cols.copy()
# define the v's for umap calculation (vs_markers_cols)
not_these = [] # columns to be excluded for umap calculation
vs_markers_cols = read_marker_csv(
input_dir) #Read them from a .csv file in ./input
print(vs_markers_cols)
df_vs_markers_cols = pd.DataFrame(vs_markers_cols, columns=['marker'])
df_vs_markers_cols.index = np.arange(1, len(df_vs_markers_cols) + 1)
input_dir) #compare_from=inputfile
#Keep only selected markers
if filter_markers:
selected_markers = read_marker_csv(input_dir)
print(compare_to.columns)
file_cols = compare_to.columns
[x for x in file_cols if x[0].isdigit()]
for x in file_cols:
if x[0].isdigit():
if x not in selected_markers:
compare_to = compare_to.drop(x, axis=1)
print(compare_to.columns)
# compare_to = downsample_data(compare_to, info_run, output_dir)
compare_to_arc, marker_list = arcsinh_transf(cofactor, compare_to)
print('Sample files:')
print('\n'.join([f for f in input_files]))
print(f'\nReference:\n{denominator}')
print('\nMarkers:')
print('\n'.join([m for m in marker_list]))
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Perform EMD~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~#
#Calculate EMD and save to the output folder (with denominator info run):
# calculate emd and sign the emd score by the difference of median between compare_from and compare_to
emd_df = pd.DataFrame()
emd_infodict = {}
emd_infodict["denominator"] = denominator