[BUG] fix: add missing max_len to GreedyEncoder.get_test_params param0 (closes #376)

pyaptamer/aptanet/_feature_classifier.py

@@ -9,7 +9,19 @@
 from sklearn.feature_selection import SelectFromModel
 from sklearn.pipeline import Pipeline
 from sklearn.utils.multiclass import type_of_target
-from sklearn.utils.validation import check_is_fitted, validate_data
+from sklearn.utils.validation import check_is_fitted
+import sklearn
+from packaging.version import Version
+
+if Version(sklearn.__version__) >= Version("1.6"):
+    from sklearn.utils.validation import validate_data
+else:
+    from sklearn.utils.validation import check_array, check_X_y
+
+    def validate_data(estimator, X, y=None, **kwargs):
+        if y is not None:
+            return estimator._validate_data(X, y, **kwargs)
+        return estimator._validate_data(X, **kwargs)
 from torch import optim
 
 from pyaptamer.aptanet._aptanet_nn import AptaNetMLP

pyaptamer/trafos/encode/_greedy.py

@@ -140,9 +140,12 @@ def get_test_params(self):
         params : dict
             Test parameters for GreedyEncoder.
         """
+
         param0 = {
             "words": {"A": 1, "C": 2, "G": 3, "U": 4, "AC": 5, "GU": 6},
+            "max_len": 10,
         }
+
         param1 = {
             "words": {"A": 1, "C": 2, "G": 3, "U": 4},
             "max_len": 10,

pyaptamer/utils/_rna.py

@@ -152,23 +152,22 @@ def rna2vec(
         ]
 
         # skip sequences that convert to an empty list
-        if any(converted):
-            # truncate if too long
-            if max_sequence_length is not None and len(converted) > max_sequence_length:
-                converted = converted[:max_sequence_length]
-
-            # pad if too short
-            if max_sequence_length is not None:
-                pad_length = max_sequence_length - len(converted)
-                padded_sequence = np.pad(
-                    array=converted,
-                    pad_width=(0, pad_length),
-                    constant_values=0,
-                )
-            else:
-                padded_sequence = np.array(converted)
-
-            result.append(padded_sequence)
+# truncate if too long
+        if max_sequence_length is not None and len(converted) > max_sequence_length:
+            converted = converted[:max_sequence_length]
+
+        # pad if too short (including empty/short sequences which become all zeros)
+        if max_sequence_length is not None:
+            pad_length = max_sequence_length - len(converted)
+            padded_sequence = np.pad(
+                array=converted,
+                pad_width=(0, pad_length),
+                constant_values=0,
+            )
+        else:
+            padded_sequence = np.array(converted)
+
+        result.append(padded_sequence)
 
     return np.array(result)