def main():
# user input
input_params = SimpleNamespace()
# Make adjustments here:
input_params.parent_path = '/Users/swilson/Documents/Grad_Folder2018/young_lab_rotation/Fiji-Images/test_data_3D'
input_params.output_path = '/Users/swilson/Documents/Grad_Folder2018/young_lab_rotation/Fiji-Images/output_3D'
file_extension = '.nd'
separate_rep_imgs = True
if not os.path.isdir(input_params.output_path):
os.mkdir(input_params.output_path)
#parse directory of data and run analysis on each replicate
data_files = methods.parse_tree(input_params.parent_path, file_extension)
npuncta_list = []
exp_list = []
nuclei_list = []
intensities_hp1 = []
intensities_med1 = []
for condition, rep_files in data_files.items():
excel_output = pd.DataFrame(columns=[
'sample', 'replicate_id', 'nuclear_id', 'total_nuc_voxels',
'channel', 'mean_in', 'mean_out', 'norm_mean', 'total_in',
'total_out', 'norm_total'
])
replicate_count = 1
for rep in rep_files: #REPLICATES
data = SimpleNamespace()
base_file = [f for f in rep if file_extension in f][0]
data.rep_name = base_file.replace(file_extension, '')
data.condition = condition
data.rep_files = rep
data = methods.load_images(data, input_params)
data = methods.find_nucleus_3D(data, input_params)
spots = pd.DataFrame(
columns=['nuc_id', 'spot_id', 'channel', 'r', 'c', 'z'])
spot_count = 0
for channel in data.pro_imgs:
print(data.rep_name)
print(channel)
### Second Attempt ###
if channel == 'ch488':
if 'HP1' in data.rep_name:
blur_img = methods.gaussian_blur(
data.pro_imgs[channel])
elif 'Med1' in data.rep_name:
blur_img = methods.gaussian_blur(
data.pro_imgs[channel])
elif 'fib1' in data.rep_name:
blur_img = methods.gaussian_blur(
data.pro_imgs[channel])
puncta_labels, npuncta, puncta, puncta_mask = methods.threshold_puncta(
blur_img, data, input_params, channel)
npuncta_list.append(npuncta)
exp = data.rep_name + "_" + channel
exp_list.append(exp)
if 'HP1' in data.rep_name:
intensities = methods.intensity_at_puncta(
puncta_labels, data.pro_imgs['ch561'], npuncta,
data, input_params)
intensities_hp1.append(intensities)
elif 'Med1' in data.rep_name:
intensities = methods.intensity_at_puncta(
puncta_labels, data.pro_imgs['ch561'], npuncta,
data, input_params)
intensities_med1.append(intensities)
manders_list = []
for i in range(1, 10):
print(i)
random_puncta_mask = methods.random_regions(
npuncta, puncta, data.pro_imgs['ch561'], data,
input_params)
manders = methods.manders(random_puncta_mask,
data.pro_imgs['ch561'],
npuncta, data, input_params)
manders_list.append(manders)
print(np.mean(manders_list))
manders_protein = methods.manders(puncta_mask,
data.pro_imgs['ch561'],
npuncta, data,
input_params)
print(manders_protein)
print('DONE')
## Colocalization analysis
## Two colors
#no_backgrnd_img_488 = methods.subtract_median(data.pro_imgs['ch488'],data,input_params)
#no_backgrnd_img_561 = methods.subtract_median(data.pro_imgs['ch561'],data,input_params)
#methods.colocalize(no_backgrnd_img_488,no_backgrnd_img_561,data,input_params)#no_backgrnd_img_488,no_backgrnd_img_561,data,input_params)
npuncta_list.append(npuncta)
exp = data.rep_name + "_" + channel
exp_list.append(exp)
f = open("num_puncta_per_channel_cisplatin.csv", 'w')
for exp in exp_list:
f.write(exp)
f.write("\t")
f.write("\n")
for n in npuncta_list:
f.write(str(n))
f.write("\t")
num_bins = 50
flattened_hp1 = [val for sublist in intensities_hp1 for val in sublist]
plt.hist(flattened_hp1, num_bins, facecolor='blue', alpha=0.5)
plt.savefig('histogram_hp1_allimgs' + '.png', dpi=300)
plt.close()
num_bins = 50
flattened_med1 = [val for sublist in intensities_med1 for val in sublist]
plt.hist(flattened_med1, num_bins, facecolor='blue', alpha=0.5)
plt.savefig("histogram_med1_allimgs" + '.png', dpi=300)
plt.close()
replicate_output = pd.DataFrame()
input_params.replicate_count = 1
for idx, file in enumerate(base_name_files):
data = SimpleNamespace(
) # this is the session data object that will be passed to functions. Corresponds to one replicate
sample_name = file.replace(file_ext, '')
replicate_files = [
os.path.join(input_params.parent_path, folder, r)
for r in file_list if sample_name in r and os.path.isfile(
os.path.join(input_params.parent_path, folder, r))
]
replicate_files = np.sort(replicate_files)
data = methods.load_images(replicate_files, data, input_params,
folder)
data = methods.find_scaffold(data, input_params)
data, rep_bulk, rep_total = methods.find_droplets(
data, input_params)
data = methods.measure_droplets(data, input_params, rep_bulk)
replicate_output = replicate_output.append(data.replicate_output,
ignore_index=True)
if len(bulk_sig) == 0:
for c in data.channel_names:
bulk_sig[c] = [rep_bulk[c]]
total_sig[c] = [rep_total[c]]
else:
for c in data.channel_names:
bulk_sig[c].append(rep_bulk[c])
if whichdataset == 'York':
input = np.load('YCMRI_128x128_images.npy')
labels = np.load('YCMRI_128x128_labels.npy')
path = 'York_results'
if whichdataset == 'ACDC':
data_dir = 'ACDC_dataset/training'
raw_image_path01 = '_frame01.nii.gz'
raw_image_path12 = '_frame12.nii.gz'
label_path01 = '_frame01_gt.nii.gz'
label_path12 = '_frame12_gt.nii.gz'
path = 'ACDC_results'
images_paths, labels_paths = methods.load_images(data_dir, raw_image_path01, raw_image_path12, label_path01, label_path12)
images_paths.sort() # each label should be on the same index than its corresponding image
labels_paths.sort()
img_data = methods.load_data(images_paths)
input = np.load('unet_input.npy')
labels = np.load('unet_labels.npy')
unet_input = []
unet_labels = []
#TODO variable amount of slices per person? (in percentages)
total_number_of_patients = len(input)
def main(data_path, nuc_idx=0, pro_idx=1, threshold=None):
# user input
input_params = SimpleNamespace()
input_params.parent_path = data_path
input_params.nuc_idx = nuc_idx
input_params.pro_idx = pro_idx
if threshold is not None:
input_params.threshold = threshold / 65536
else:
input_params.threshold = threshold
folder_list = os.listdir(input_params.parent_path)
folder_list.sort(reverse=False)
file_ext = '.czi'
# make output directories
#input_params.output_path = input_params.parent_path
input_params.output_path = '/lab/solexa_young/scratch/MECP2_Imaging/20191112_neuron_imaging/Volumes_extra/gaussian-sigma3_mean2.35_struct3_dist0.2'
if not os.path.isdir(input_params.output_path):
os.mkdir(input_params.output_path)
for folder in folder_list:
if not folder.startswith('.') and not folder.endswith(
'output') and os.path.isdir(
os.path.join(input_params.parent_path,
folder)): #SAMPLES/EXPERIMENTS
print()
print('Started: ', folder, ' at ', datetime.now())
print()
temp_output = os.path.join(input_params.output_path,
folder + '_output')
if not os.path.isdir(temp_output):
os.mkdir(temp_output)
file_list = os.listdir(
os.path.join(input_params.parent_path, folder))
base_name_files = [
f for f in file_list if file_ext in f and os.path.isfile(
os.path.join(input_params.parent_path, folder, f))
]
base_name_files.sort(reverse=False)
excel_output = pd.DataFrame(columns=[
'sample', 'replicate_id', 'nuc_id', 'total_nuc_voxels',
'channel', 'mean_in', 'mean_out', 'norm_mean', 'total_in',
'total_out', 'norm_total'
])
objects_output = pd.DataFrame(columns=[
'sample', 'replicate_id', 'nuc_id', 'object_id', 'voxels',
'channel', 'mean_in', 'mean_out', 'norm_mean'
])
replicate_count = 1
for idx, file in enumerate(base_name_files): #REPLICATES
print()
print(file)
print()
data = SimpleNamespace()
data.sample_name = file.replace(file_ext, '')
data.folder = folder
data.img_path = os.path.join(input_params.parent_path, folder,
file)
data = methods.load_images(data, input_params)
if data is not None:
data = methods.find_nucleus_2D(data, input_params)
data.z_count = data.nuc_img.shape[0]
if idx == 0:
# z = int(data.nucleus_image.shape[0]/2)
z = 10
# make_nucleus_montage_2D(data, input_params)
total_dense_object_mask = np.full(shape=data.nuc_img.shape,
fill_value=False,
dtype=bool)
for r_idx, region in enumerate(data.nuc_regions):
region_area = methods.find_region_area(region)
if region_area >= 30000:
nuc_id = data.nuc_label[int(
(region[0].stop + region[0].start) / 2),
int((region[1].stop +
region[1].start) / 2)]
nuc_box = data.nuc_img[:, region[0], region[1]]
nuc_mask_box = data.nuc_label[region[0], region[1]]
single_nuc_mask = nuc_mask_box == nuc_id
single_nuc_mask = np.repeat(
single_nuc_mask[np.newaxis, :, :],
data.z_count,
axis=0
) # because our nuclear mask is 2D so we project it to 3D
dense_obj_mask, bg_nuc_mask, dense_objects = methods.find_dense_objects_3D(
nuc_box, single_nuc_mask, input_params, data)
total_dense_object_mask[:, region[0], region[1]][
dense_obj_mask] = True
for p_idx, image in enumerate(data.pro_imgs):
channel_name = data.pro_ch_names[p_idx]
protein_box = image[:, region[0], region[1]]
mean_in = np.mean(protein_box[dense_obj_mask])
total_in = np.sum(protein_box[dense_obj_mask])
mean_out = np.mean(protein_box[bg_nuc_mask])
total_out = total_in + np.sum(
protein_box[bg_nuc_mask])
norm_mean = mean_in / mean_out
norm_total = total_in / total_out
nuc_voxels = np.sum(dense_obj_mask)
excel_output = excel_output.append(
{
'sample': folder,
'replicate_id': replicate_count,
'nuc_id': nuc_id,
'total_voxels': nuc_voxels,
'channel': str(channel_name),
'mean_in': mean_in,
'mean_out': mean_out,
'norm_mean': norm_mean,
'total_in': total_in,
'total_out': total_out,
'norm_total': norm_total
},
ignore_index=True)
for o_idx, object in enumerate(dense_objects):
voxels = np.sum(dense_obj_mask[object])
mean_in = np.mean(
protein_box[object]
) # not perfect because this is just a 3D bounding box, which will include pixels not in the region, but good enough for now!
objects_output = objects_output.append(
{
'sample': folder,
'replicate_id': replicate_count,
'nuc_id': nuc_id,
'object_id': o_idx + 1,
'voxels': voxels,
'channel': str(channel_name),
'mean_in': mean_in,
'mean_out': mean_out,
'norm_mean': mean_in / mean_out
},
ignore_index=True)
graph_output_path = os.path.join(
temp_output,
folder + '_rep' + str(replicate_count) + '.png')
methods.make_output_graphs(data.nuc_label,
total_dense_object_mask, data,
graph_output_path)
replicate_count += 1
else:
replicate_count += 1
excel_output.to_excel(os.path.join(temp_output,
folder + '_enrichment.xlsx'),
index=False)
objects_output.to_excel(os.path.join(temp_output,
folder + '_objects.xlsx'),
index=False)