Coverage for src/abm_initialization_collection/image/select_fov_images.py: 100%

1import numpy as np 

2import pandas as pd 

3 

4 

5def select_fov_images( 

6 metadata: pd.DataFrame, cells_per_fov: int, bins: list[int], counts: list[int] 

7) -> list[dict]: 

8 """ 

9 Selects FOV images from dataset within each volume bin. 

10 

11 Selected FOVs must not contains any outliers or cells in a state other than 

12 M0. Selected FOVs will have exactly the number of specified cells. Average 

13 volume of all cells in the FOV are used to determine which volume bin the 

14 FOV falls in. FOVs are evaluated in the order they appear in the metadata. 

15 

16 Parameters 

17 ---------- 

18 metadata 

19 Quilt package metadata for FOV images. 

20 cells_per_fov 

21 Number of cells per FOV. 

22 bins 

23 Cell volume bin boundaries. 

24 counts 

25 Number of FOVs to select from each cell volume bin. 

26 

27 Returns 

28 ------- 

29 : 

30 List of selected FOV image paths and cell ids 

31 """ 

32 

33 total_count = sum(counts) 

34 bin_counts = [0] * len(counts) 

35 fovs = [] 

36 

37 for fov_seg_path, seg_group in metadata.groupby("fov_seg_path"): 

38 no_outliers = (seg_group.outlier != "Yes").all() 

39 only_m0 = (seg_group.cell_stage == "M0").all() 

40 

41 if no_outliers and only_m0 and len(seg_group) == cells_per_fov: 

42 mean = seg_group["MEM_shape_volume"].mean() 

43 bin_index = np.digitize(mean, bins) - 1 

44 

45 if bin_counts[bin_index] < counts[bin_index]: 

46 bin_counts[bin_index] = bin_counts[bin_index] + 1 

47 key = fov_seg_path.split("/")[-1].split("_")[0] 

48 fovs.append( 

49 { 

50 "key": key, 

51 "item": fov_seg_path, 

52 "cell_ids": list(seg_group["this_cell_index"]), 

53 } 

54 ) 

55 

56 if sum(bin_counts) >= total_count: 

57 break 

58 

59 return fovs