cc-mapping 0.2.0__tar.gz → 0.2.2__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/PKG-INFO +1 -1
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/pyproject.toml +1 -1
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/thresholding/sequential.py +21 -12
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/LICENSE +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/README.md +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/__init__.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/core.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/feature_selection/__init__.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/feature_selection/_base.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/feature_selection/_persistence.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/feature_selection/_plotting.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/feature_selection/_preprocessing.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/feature_selection/_random_forest.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/feature_selection/_training.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/manifold.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/plot.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/preprocess.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/pseudotime.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/thresholding/__init__.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/thresholding/base.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/thresholding/single.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/utils.py +0 -0
- {cc_mapping-0.2.0 → cc_mapping-0.2.2}/src/cc_mapping/utils_v0.py +0 -0
|
@@ -429,10 +429,14 @@ class SequentialGMM(GaussianMixtureModelBase):
|
|
|
429
429
|
data_probs=gmm.predict_proba(subset_data.reshape(-1, 1)),
|
|
430
430
|
)
|
|
431
431
|
|
|
432
|
-
# Handle duplicate labels if needed
|
|
433
|
-
|
|
434
|
-
|
|
435
|
-
|
|
432
|
+
# Handle duplicate labels if needed.
|
|
433
|
+
# Match single.py:643 arg order: (data_probs, ordered_labels) — and only
|
|
434
|
+
# collapse when labels actually contain duplicates, since the underlying
|
|
435
|
+
# _handle_duplicate_labels can mis-shape the output for no-op collapses.
|
|
436
|
+
has_duplicates = len(set(ordered_labels)) < len(ordered_labels)
|
|
437
|
+
if duplicate_labels and has_duplicates:
|
|
438
|
+
condensed_data_probs, ordered_labels_processed = self._handle_duplicate_labels(
|
|
439
|
+
gmm_info.data_probs, ordered_labels
|
|
436
440
|
)
|
|
437
441
|
gmm_info.condensed_data_probs = condensed_data_probs
|
|
438
442
|
else:
|
|
@@ -440,19 +444,24 @@ class SequentialGMM(GaussianMixtureModelBase):
|
|
|
440
444
|
|
|
441
445
|
# Calculate decision boundaries from probabilities
|
|
442
446
|
# Convert data_probs back to numpy array (Pydantic stores as list)
|
|
443
|
-
probs_array = np.array(gmm_info.condensed_data_probs if duplicate_labels else gmm_info.data_probs)
|
|
447
|
+
probs_array = np.array(gmm_info.condensed_data_probs if (duplicate_labels and has_duplicates) else gmm_info.data_probs)
|
|
444
448
|
decision_boundaries = self._calculate_decision_boundaries_from_probs(
|
|
445
449
|
feature_values=subset_data,
|
|
446
450
|
data_probs=probs_array,
|
|
447
451
|
ordered_labels=ordered_labels_processed,
|
|
448
452
|
)
|
|
449
453
|
|
|
450
|
-
# Assign new labels
|
|
451
|
-
|
|
452
|
-
|
|
453
|
-
|
|
454
|
-
|
|
455
|
-
|
|
454
|
+
# Assign new labels via per-cell argmax of the (possibly condensed)
|
|
455
|
+
# probability matrix. This is robust to overlapping components: when
|
|
456
|
+
# the threshold-derivation walk produces fewer than N-1 forward
|
|
457
|
+
# transitions (because backward transitions in the sorted-cell
|
|
458
|
+
# sequence get filtered), label assignment used to error out with
|
|
459
|
+
# "labels must equal thresholds + 1". decision_boundaries are still
|
|
460
|
+
# computed above and stored in internal_data for plotting / .uns
|
|
461
|
+
# introspection; only the cell-label assignment switches to argmax.
|
|
462
|
+
class_idx = np.argmax(probs_array, axis=1)
|
|
463
|
+
labels_arr = np.asarray(ordered_labels_processed)
|
|
464
|
+
new_labels = labels_arr[class_idx]
|
|
456
465
|
|
|
457
466
|
# Update labels in-place
|
|
458
467
|
self._update_labels_in_place(
|
|
@@ -469,7 +478,7 @@ class SequentialGMM(GaussianMixtureModelBase):
|
|
|
469
478
|
gmm_info=gmm_info,
|
|
470
479
|
ordered_gmm_labels=ordered_labels,
|
|
471
480
|
decision_boundaries=decision_boundaries,
|
|
472
|
-
condensed_labels=ordered_labels_processed if duplicate_labels else None,
|
|
481
|
+
condensed_labels=ordered_labels_processed if (duplicate_labels and has_duplicates) else None,
|
|
473
482
|
feature_name=feature,
|
|
474
483
|
gmm_obs_label=obs_label,
|
|
475
484
|
)
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|