msasim 2024.5.22__cp312-cp312-musllinux_1_1_x86_64.whl → 2024.5.2720__cp312-cp312-musllinux_1_1_x86_64.whl

Sailfish/simulator.py → msasim/sailfish.py

@@ -368,8 +368,8 @@ class Msa:
     '''
     The MSA class from the simulator
     '''
-    def __init__(self, species_dict: Dict[str, BlockTree], root_node):
-        self._msa = _Sailfish.Msa(species_dict, root_node)
+    def __init__(self, species_dict: Dict[str, BlockTree], root_node, save_list: List[bool]):
+        self._msa = _Sailfish.Msa(species_dict, root_node, save_list)
     
     def generate_msas(self, node):
         self._msa.generate_msas(node)
@@ -504,7 +504,15 @@ class Simulator:
     def gen_indels(self) -> BlockTreePython:
         return BlockTreePython(self._simulator.gen_indels())
     
+    def get_sequences_to_save(self) -> List[bool]:
+        return self._simulator.get_saved_nodes_mask()
     
+    def save_root_sequence(self):
+        self._simulator.save_root_sequence()
+    
+    def save_all_nodes_sequences(self):
+        self._simulator.save_all_nodes_sequences()
+
     def gen_substitutions(self, length: int):
         if not self._is_sub_model_init:
             self._init_sub_model()
@@ -515,10 +523,14 @@ class Simulator:
         Msas = []
         for _ in range(times):
             if self._simProtocol._is_insertion_rate_zero and self._simProtocol._is_deletion_rate_zero:
-                msa = Msa(self._simProtocol.get_tree().get_num_leaves(), self._simProtocol.get_sequence_size())
+                msa = Msa(sum(self.get_sequences_to_save()),
+                          self._simProtocol.get_sequence_size(),
+                          self.get_sequences_to_save())
             else:
                 blocktree = self.gen_indels()
-                msa = Msa(blocktree._get_Sailfish_blocks(), self._simProtocol._get_root())
+                msa = Msa(blocktree._get_Sailfish_blocks(),
+                          self._simProtocol._get_root(),
+                          self.get_sequences_to_save())
 
             # sim.init_substitution_sim(mFac)
             substitutions = self.gen_substitutions(msa.get_length())

msasim-2024.5.2720.dist-info/METADATA

@@ -0,0 +1,54 @@
+Metadata-Version: 2.1
+Name: msasim
+Version: 2024.5.2720
+Summary: A fast MSA simulator
+Home-page: https://github.com/elyawy/Sailfish-backend
+Author: Elya Wygoda
+Author-email: elya.wygoda@gmail.com
+Requires-Python: >=3.6
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Provides-Extra: test
+Requires-Dist: pytest ; extra == 'test'
+
+# Sailfish
+
+Sailfish is a performant multiple sequence alignment(MSA) simulator, written in C++ and Python, allowing for quick and easy generation of large simulated datasets. 
+
+## Project goals
+
+- Ease of use
+- Speed
+- Modularity
+
+## Installation
+
+```bash
+pip install msasim
+```
+
+## Example
+
+```python
+from msasim import sailfish as sim
+from msasim.sailfish import MODEL_CODES, ZipfDistribution
+
+ROOT_SEQUENCE_LENGTH = 100
+
+sim_protocol = sim.SimProtocol("(A:0.5,B:0.5);",
+                               deletion_rate=0.01,
+                               insertion_rate=0.01,
+                               deletion_dist=ZipfDistribution(1.08, 50),
+                               insertion_dist=ZipfDistribution(1.08, 50),
+                               seed=50)
+sim_protocol.set_sequence_size(ROOT_SEQUENCE_LENGTH)
+
+simulation = sim.Simulator(sim_protocol, simulation_type=sim.SIMULATION_TYPE.PROTEIN)
+
+simulation.set_replacement_model(model=MODEL_CODES.WAG, 
+                                 gamma_parameters_alpha=1.0, 
+                                 gamma_parameters_catergories=4)
+msa = simulation()
+msa.print_msa()
+
+```

msasim-2024.5.2720.dist-info/RECORD

@@ -0,0 +1,10 @@
+_Sailfish.cpython-312-x86_64-linux-musl.so,sha256=hbZ3mj5XwQX_X64AgKuDBA0tyYpGe8S5my8_8o4NfJA,1193009
+msasim-2024.5.2720.dist-info/WHEEL,sha256=u3WJvwcnUGXBzVB5YEzh4MIq4CvlLxbaEALwCGrbBH8,113
+msasim-2024.5.2720.dist-info/LICENSE,sha256=p1Aa_mM2Nu6dG3XqMVU62Jhf04lNOwtXUrhhvhcDips,10312
+msasim-2024.5.2720.dist-info/top_level.txt,sha256=NS1ILx5V94Yyh_M7yDrrqbBPu1TL_zJuxhsI90YEJVY,17
+msasim-2024.5.2720.dist-info/RECORD,,
+msasim-2024.5.2720.dist-info/METADATA,sha256=sTra7J_hCPNlhMvIMhL0ASj-HqGczjjoNPayjJhRWH8,1472
+msasim.libs/libstdc++-a9383cce.so.6.0.28,sha256=Wy9UCdwS1rwI9GU5e7qE61S0AkRqqwti1q_adWSs-Rk,2447393
+msasim.libs/libgcc_s-a04fdf82.so.1,sha256=YxqJNaesQMhDswHEQpXsiLnVvMBBbYO6KYMDZFPWKSM,81257
+msasim/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+msasim/sailfish.py,sha256=zLW65fPWCwOUDpVyq4zdZK2YsGcwMUZxnUVzeRnFVbo,24791

{msasim-2024.5.22.dist-info → msasim-2024.5.2720.dist-info}/top_level.txt

@@ -1,2 +1,2 @@
-Sailfish
 _Sailfish
+msasim

msasim-2024.5.22.dist-info/METADATA

@@ -1,13 +0,0 @@
-Metadata-Version: 2.1
-Name: msasim
-Version: 2024.5.22
-Summary: A fast MSA simulator
-Home-page: https://github.com/elyawy/Sailfish-backend
-Author: Elya Wygoda
-Author-email: elya.wygoda@gmail.com
-Requires-Python: >=3.6
-License-File: LICENSE
-Provides-Extra: test
-Requires-Dist: pytest ; extra == 'test'
-
-Sailfish is a performant multiple sequence alignment simulator, written in C++, allowing fast generation of large simualted datasets.

msasim-2024.5.22.dist-info/RECORD

@@ -1,10 +0,0 @@
-_Sailfish.cpython-312-x86_64-linux-musl.so,sha256=_FHYmdOGc4RZA9HXppBuHFD9Lqhm8XShyrbq4ru2b68,1165969
-Sailfish/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-Sailfish/simulator.py,sha256=CTFAnavIHy918aF141FKeZjO5Y-oYASI-TR5MMGGVks,24324
-msasim.libs/libstdc++-a9383cce.so.6.0.28,sha256=Wy9UCdwS1rwI9GU5e7qE61S0AkRqqwti1q_adWSs-Rk,2447393
-msasim.libs/libgcc_s-a04fdf82.so.1,sha256=YxqJNaesQMhDswHEQpXsiLnVvMBBbYO6KYMDZFPWKSM,81257
-msasim-2024.5.22.dist-info/WHEEL,sha256=u3WJvwcnUGXBzVB5YEzh4MIq4CvlLxbaEALwCGrbBH8,113
-msasim-2024.5.22.dist-info/LICENSE,sha256=p1Aa_mM2Nu6dG3XqMVU62Jhf04lNOwtXUrhhvhcDips,10312
-msasim-2024.5.22.dist-info/top_level.txt,sha256=mVJT9Q6vqdBJyJ6acJvUdopsnVCQH7wEMQ-3iT6C-Jg,19
-msasim-2024.5.22.dist-info/RECORD,,
-msasim-2024.5.22.dist-info/METADATA,sha256=xZjkEbQzNYYvB1WB75UQGh14FnS4rM7CUgujkplkRkA,435