geney 1.4.22__py2.py3-none-any.whl → 1.4.24__py2.py3-none-any.whl

geney/utils/SeqMats.py

@@ -8,7 +8,7 @@ from typing import List, Tuple, Union, Optional
 from collections import defaultdict
 import numpy as np
 import pandas as pd
-
+from geney.utils.utils import contains
 
 
 def format_mut_id(text):
@@ -105,7 +105,7 @@ class SeqMat:
             "name": self.name,
             "source": self.source,
             "version": self.version,
-            "notes": self.notes
+            "notes": self.notes,
         }
 
     @property
@@ -159,38 +159,77 @@ class SeqMat:
         new.seq_array['valid_mask'] = new.seq_array['nt'] != b'-'
         return new
 
+    # def apply_mutations(
+    #         self,
+    #         mutations: Union[Tuple[float, str, str], List[Tuple[float, str, str]]],
+    #         only_snps: bool = False
+    # ) -> SeqMat:
+
     def apply_mutations(
             self,
-            mutations: Union[Tuple[float, str, str], List[Tuple[float, str, str]]],
+            mutations: Union[Tuple[float, str, str], List[Tuple[float, str, str]]] = None,
+            *,
+            pos: Optional[float] = None,
+            ref: Optional[str] = None,
+            alt: Optional[str] = None,
             only_snps: bool = False
-    ) -> SeqMat:
+        ) -> SeqMat:
         """
         Apply one or a batch of mutations (pos, ref, alt) efficiently:
         - Supports a single tuple or a list of tuples
         - Assumes mutations sorted by position for vectorized searchsorted
         """
+        turn_back = False
+        if self.rev:
+            turn_back = True
+            self.reverse_complement()
+
         # Normalize to list
-        if isinstance(mutations, tuple) and len(mutations) == 3:
+        # if isinstance(mutations, tuple) and len(mutations) == 3:
+        #     mutations = [mutations]
+        # elif not isinstance(mutations, list):
+        #     raise TypeError("mutations must be a tuple or list of tuples")
+        # Input normalization
+        if mutations is None:
+            if pos is None or ref is None or alt is None:
+                raise ValueError("Either `mutations` or `pos, ref, alt` must be provided")
+            mutations = [(pos, ref, alt)]
+        elif isinstance(mutations, tuple) and len(mutations) == 3:
             mutations = [mutations]
         elif not isinstance(mutations, list):
-            raise TypeError("mutations must be a tuple or list of tuples")
-
-        # Left-normalize and bucket
+            raise TypeError("`mutations` must be a tuple or list of tuples")
+
+        # # Left-normalize and bucket
+        # subs, ins, dels = [], [], []
+        # for pos, ref, alt in mutations:
+        #     while ref and alt and ref[0] == alt[0]:
+        #         pos += 1
+        #         ref = ref[1:] or '-'
+        #         alt = alt[1:] or '-'
+        #     if ref != '-' and alt != '-':
+        #         subs.append((pos, ref, alt))
+        #     elif ref == '-' and alt != '-' and not only_snps:
+        #         ins.append((pos, alt))
+        #     elif alt == '-' and ref != '-' and not only_snps:
+        #         dels.append((pos, ref))
+        #     else:
+        #         raise ValueError(f"Unsupported mutation {pos}:{ref}:{alt}.")
+        # Bucket mutations
         subs, ins, dels = [], [], []
-        for pos, ref, alt in mutations:
-            while ref and alt and ref[0] == alt[0]:
-                pos += 1
-                ref = ref[1:] or '-'
-                alt = alt[1:] or '-'
-            if ref != '-' and alt != '-':
-                subs.append((pos, ref, alt))
-            elif ref == '-' and alt != '-' and not only_snps:
-                ins.append((pos, alt))
-            elif alt == '-' and ref != '-' and not only_snps:
-                dels.append((pos, ref))
-            else:
-                raise ValueError(f"Unsupported mutation {pos}:{ref}:{alt}.")
+        for p, r, a in mutations:
+            # left-normalize
+            while r and a and r[0] == a[0]:
+                p += 1; r = r[1:] or '-'; a = a[1:] or '-'
+
+            if not contains(self.index, p):
+                continue  # Skip invalid positions
 
+            if r != '-' and a != '-': subs.append((p, r, a))
+            elif r == '-' and a != '-' and not only_snps: ins.append((p, a))
+            elif a == '-' and r != '-' and not only_snps: dels.append((p, r))
+            else: raise ValueError(f"Unsupported mutation {p}:{r}:{a}")
+
+        applied_mutations = 0
         # Ensure seq_array indices sorted
         coords = self.seq_array['index']
 
@@ -205,7 +244,7 @@ class SeqMat:
                     actual = self.seq_array['ref'][idx:idx + length].tobytes().decode()
                     raise ValueError(f"Ref mismatch at {pos}: expected {ref}, found {actual}")
                 self.seq_array['nt'][idx:idx + length] = np.frombuffer(alt.encode(), dtype='S1')
-
+                applied_mutations += 1
         # 2) Bulk insertions
         if ins:
             ins.sort(key=lambda x: x[0])
@@ -219,6 +258,8 @@ class SeqMat:
                     new_idx = pos + (cnt + i + 1) * eps
                     new_rows.append((nt.encode(), new_idx, b'-', np.nan, True))
                 self.insertion_counters[pos] += len(alt)
+                applied_mutations += 1
+
             merged = np.concatenate([self.seq_array, np.array(new_rows, dtype=self.seq_array.dtype)])
             merged.sort(order='index')
             self.seq_array = merged
@@ -232,10 +273,17 @@ class SeqMat:
             for (pos, ref), idx in zip(dels, idxs):
                 length = len(ref)
                 mask[idx:idx + length] = False
+                applied_mutations += 1
+
             self.seq_array = self.seq_array[mask]
 
         # Finalize valid mask
         self.seq_array['valid_mask'] = self.seq_array['nt'] != b'-'
+
+        if turn_back:
+            self.reverse_complement()
+
+        self.notes['total_mutations'] = self.notes.get('total_mutations', 0) + applied_mutations
         return self
 
     def complement(self) -> SeqMat:

{geney-1.4.22.dist-info → geney-1.4.24.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: geney
-Version: 1.4.22
+Version: 1.4.24
 Summary: A Python package for gene expression modeling.
 Home-page: https://github.com/nicolaslynn/geney
 Author: Nicolas Lynn

{geney-1.4.22.dist-info → geney-1.4.24.dist-info}/RECORD

@@ -37,7 +37,7 @@ geney/translation_initiation/tis_utils.py,sha256=AF3siFjuQH-Rs44EV-80zHdbxRMvN4w
 geney/translation_initiation/resources/kozak_pssm.json,sha256=pcd0Olziutq-6H3mFWDCD9cujQ_AlZO-iiOvBl82hqE,1165
 geney/translation_initiation/resources/tis_regressor_model.joblib,sha256=IXb4DUDhJ5rBDKcqMk9zE3ECTZZcdj7Jixz3KpoZ7OA,2592025
 geney/utils/Fasta_segment.py,sha256=weB5NJ65P0XiyAJCiCHx4T9sHC1pWLpuQeOy0B85gyg,11364
-geney/utils/SeqMats.py,sha256=Hneqxz92WFrHi0lyHs2ZwTd091TtFclgybcvtUCktJA,15689
+geney/utils/SeqMats.py,sha256=PaUp6PMKYDYZ8RTodmKAmCa9ywHnkqSEqTjPoATr82k,17616
 geney/utils/SeqMatsOld.py,sha256=syRU5DAuTh3xUfGW_qP9wlcBO5pHsG_y5PlrfXTIxUY,18502
 geney/utils/TranscriptLibrary.py,sha256=ma_ZVPgglxXDDneEvdqxxeqxG8eSFL-zgLUXyC6BqY8,2070
 geney/utils/__init__.py,sha256=-nJ-DMx1JzP-ZCe_QuQCeM0ZYIT_16jxoXDhUaO_4Oc,714
@@ -46,7 +46,7 @@ geney/utils/pangolin_utils.py,sha256=JQSPbWxdzqGFYfWQktkfLMaMSGR28eGQhNzO7MLMe5M
 geney/utils/spliceai_utils.py,sha256=VtrIbjyQxk_3lw86eWjftRYyal9OzxArJ0GV5u_ymTg,2721
 geney/utils/splicing_utils.py,sha256=vPCGnCPR1ooEZEHR79yFHLmRQXEJHXEQjjxpBR-YWOs,20635
 geney/utils/utils.py,sha256=m51Vd0cEbrcIHo6_8BAuI9YSPcKRs22e5LfVd2Qj6Is,2181
-geney-1.4.22.dist-info/METADATA,sha256=OH6exXPW8_IdusLX5g-xeLBXyyQx1DQyfZeguHjvyQY,990
-geney-1.4.22.dist-info/WHEEL,sha256=AHX6tWk3qWuce7vKLrj7lnulVHEdWoltgauo8bgCXgU,109
-geney-1.4.22.dist-info/top_level.txt,sha256=O-FuNUMb5fn9dhZ-dYCgF0aZtfi1EslMstnzhc5IIVo,6
-geney-1.4.22.dist-info/RECORD,,
+geney-1.4.24.dist-info/METADATA,sha256=myrl1R2vXV7P8TYZX6rUP_xTDFfNBE9vrv2yAzUFpEM,990
+geney-1.4.24.dist-info/WHEEL,sha256=AHX6tWk3qWuce7vKLrj7lnulVHEdWoltgauo8bgCXgU,109
+geney-1.4.24.dist-info/top_level.txt,sha256=O-FuNUMb5fn9dhZ-dYCgF0aZtfi1EslMstnzhc5IIVo,6
+geney-1.4.24.dist-info/RECORD,,