Update contour_utils.py #25
Conversation
added get_morphology_PCA function
Using pre-commit and added function description
| Returns: | ||
| List: PCA values for contours | ||
| """ | ||
| flat_contour_entire_dataset = [] |
There was a problem hiding this comment.
What does it contain and what is its shape?
| traj = trajectory_collection.get_trajectory(track_id_num) | ||
|
|
||
| # getting cell contours using active shape model | ||
| cell_contours = get_cellTool_contour_points(traj, contour_num_points=contour_num_points) |
| Args: | ||
| trajectory_collection (obj): trajectory collection of a dataset | ||
| contour_num_points (int): number of landmark contour points | ||
| trajectory_threshold (int, optional): _description_. Defaults to 1. |
There was a problem hiding this comment.
what is trajectory threshold? threshold of the length of the trajectories? Users can filter via trajectory collection outside of this function. Not need to add and create high coupling logics.
There was a problem hiding this comment.
do you prefer if we create a line of code outside before runing pca?
| final_contour = np.hstack(flattened_contour) | ||
| flat_contour_entire_dataset.append(final_contour) | ||
|
|
||
| countour_entire_dataset = np.concatenate(flat_contour_entire_dataset, axis=0).reshape(-1, 2 * contour_num_points) |
There was a problem hiding this comment.
countour_entire_dataset -> all_sctc_contours
FYI
sctc - single cell trajectory collection
sct - single cell trajectory
I will add these conventions to dev details
| # getting PCA | ||
| pca = PCA(n_components=num_components, svd_solver="full") | ||
|
|
||
| pca_entire = pca.fit_transform(countour_entire_dataset) |
There was a problem hiding this comment.
pca_entire ->transformed_pca_contour_data
There was a problem hiding this comment.
and this variable is not used by other parts/return value?
There was a problem hiding this comment.
I actually wanted to talk to you more about this. I did talk to Weikang about this and maybe I can explain why he did that here in his main pipeline.
| countour_entire_dataset = np.concatenate(flat_contour_entire_dataset, axis=0).reshape(-1, 2 * contour_num_points) | ||
|
|
||
| # getting PCA | ||
| pca = PCA(n_components=num_components, svd_solver="full") |
|
|
||
| pca_individual_list = [ | ||
| pca.transform( | ||
| final_contour.reshape((int(len(final_contour) / (2 * contour_num_points)), 2 * contour_num_points)) |
There was a problem hiding this comment.
Please make code more readable
E.g. someVar = (int(len(final_contour) / (2 * contour_num_points))
I need to understand the code within 10 seconds of reading.
What is pca_individual list? It can mean a lot of things, e.g. each PCA principle components, contours after PCA transforms for each trajectory, etc. Improve.
added sct and sctc feature collection along with pca function for sctc features
Adding norm_haralick features
added get_morphology_PCA function