how to handle empty tiles

[mkunst23]

Hi,

the images from my spatial transcriptomics data don't fill the entire image so we I split up the image into smaller tiles I get an error that there are no cells (see error message below). Is there a way to instead of teminating the for loop to just skip empty tiles?

ValueError Traceback (most recent call last)
File :15, in

File /allen/programs/celltypes/workgroups/rnaseqanalysis/mFISH/michaelkunst/python/ClusterMap/ClusterMap/clustermap.py:87, in ClusterMap.preprocess(self, dapi_grid_interval, LOF, contamination, pct_filter)
86 def preprocess(self,dapi_grid_interval=5, LOF=False, contamination=0.1, pct_filter=0.1):
---> 87 preprocessing_data(self.spots, dapi_grid_interval, self.dapi_binary, LOF,contamination, self.xy_radius,pct_filter)

File /allen/programs/celltypes/workgroups/rnaseqanalysis/mFISH/michaelkunst/python/ClusterMap/ClusterMap/preprocessing.py:136, in preprocessing_data(spots, dapi_grid_interval, dapi_binary, LOF, contamination, xy_radius, pct_filter)
134 #compute neighbors within radius for local density
135 knn = NearestNeighbors(radius=xy_radius)
--> 136 knn.fit(all_points)
137 spots_array = np.array(spots.loc[:, ['spot_location_2', 'spot_location_1']])
138 neigh_dist, neigh_array = knn.radius_neighbors(spots_array)

File /allen/programs/celltypes/workgroups/rnaseqanalysis/mFISH/michaelkunst/miniconda3/envs/clustermap_hpc/lib/python3.8/site-packages/sklearn/neighbors/_unsupervised.py:166, in NearestNeighbors.fit(self, X, y)
149 def fit(self, X, y=None):
150 """Fit the nearest neighbors estimator from the training dataset.
151
152 Parameters
(...)
164 The fitted nearest neighbors estimator.
165 """
--> 166 return self._fit(X)

File /allen/programs/celltypes/workgroups/rnaseqanalysis/mFISH/michaelkunst/miniconda3/envs/clustermap_hpc/lib/python3.8/site-packages/sklearn/neighbors/_base.py:435, in NeighborsBase._fit(self, X, y)
433 else:
434 if not isinstance(X, (KDTree, BallTree, NeighborsBase)):
--> 435 X = self._validate_data(X, accept_sparse="csr")
437 self._check_algorithm_metric()
438 if self.metric_params is None:

File /allen/programs/celltypes/workgroups/rnaseqanalysis/mFISH/michaelkunst/miniconda3/envs/clustermap_hpc/lib/python3.8/site-packages/sklearn/base.py:561, in BaseEstimator._validate_data(self, X, y, reset, validate_separately, **check_params)
559 raise ValueError("Validation should be done on X, y or both.")
560 elif not no_val_X and no_val_y:
--> 561 X = check_array(X, **check_params)
562 out = X
563 elif no_val_X and not no_val_y:

File /allen/programs/celltypes/workgroups/rnaseqanalysis/mFISH/michaelkunst/miniconda3/envs/clustermap_hpc/lib/python3.8/site-packages/sklearn/utils/validation.py:797, in check_array(array, accept_sparse, accept_large_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, ensure_min_samples, ensure_min_features, estimator)
795 n_samples = _num_samples(array)
796 if n_samples < ensure_min_samples:
--> 797 raise ValueError(
798 "Found array with %d sample(s) (shape=%s) while a"
799 " minimum of %d is required%s."
800 % (n_samples, array.shape, ensure_min_samples, context)
801 )
803 if ensure_min_features > 0 and array.ndim == 2:
804 n_features = array.shape[1]

ValueError: Found array with 0 sample(s) (shape=(0, 2)) while a minimum of 1 is required.

[rocketeer1998]

I recently resolved this error by adding the below code to let the for loop skip processing the empty tiles.

for tile_num in range(out.shape[0]):
    if tile_num == 6 or tile_num == 7 or tile_num == 15 : 
        continue
    print(f'tile: {tile_num}')
    ...

where tile 6,7 and 15 are empty. Hopefully it helps.

[mkunst23]

Yes that helped. Thanks

[yichunher]

Hi,

Sorry for the late reply. Thank @rocketeer1998 for helping! I also skipped tiles with less than 20 spots.

if spots_tile.shape[0] < 20:
continue