msasim 24.10.4__tar.gz → 24.12.0__tar.gz

{msasim-24.10.4 → msasim-24.12.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: msasim
-Version: 24.10.4
+Version: 24.12.0
 Summary: A fast MSA simulator
 Home-page: https://github.com/elyawy/Sailfish-backend
 Author: Elya Wygoda

{msasim-24.10.4 → msasim-24.12.0}/msasim/sailfish.py

@@ -1,5 +1,5 @@
 import _Sailfish
-import os, warnings, math, operator, time, profile
+import os, warnings, math, operator, time, profile, tempfile, pathlib
 from functools import reduce
 from typing import List, Optional, Dict
 from re import split
@@ -101,12 +101,9 @@ class ZipfDistribution(Distribution):
         """
         self.p = p
         self.truncation = truncation
-        HARMONIC = lambda n,s: sum([(i**-s) for i in range(1,n+1)])
-        PMF = lambda x: (x**-p)*(1.0/HARMONIC(truncation, p))
-        CDF = lambda x: HARMONIC(x, p) / HARMONIC(truncation, p)
-        norm_factor = CDF(truncation) - CDF(0)
 
-        probabilities = [PMF(i)/norm_factor for i in range(1, truncation+1)]
+        norm_factor = sum([(i**-p) for i in range(1,truncation+1)])
+        probabilities = [(i**-p)/norm_factor for i in range(1, truncation+1)]
 
         self.set_dist(probabilities)
     
@@ -537,6 +534,26 @@ class Simulator:
             Msas.append(msa)
         return Msas
     
+    def simulate_low_memory(self, output_file_path: pathlib.Path) -> Msa:
+        if self._simProtocol._is_insertion_rate_zero and self._simProtocol._is_deletion_rate_zero:
+            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(),
+                        self.get_sequences_to_save())
+
+        # sim.init_substitution_sim(mFac)
+        if self._simulation_type != SIMULATION_TYPE.NOSUBS:
+            with tempfile.TemporaryDirectory() as tmpdirname:
+                self._simulator.gen_substitutions_to_dir(msa.get_length(), tmpdirname)
+                msa._msa.set_substitutions_folder(tmpdirname)
+                msa._msa.write_msa_from_dir(str(output_file_path))
+
+
+    
     def __call__(self) -> Msa:
         return self.simulate(1)[0]
     

{msasim-24.10.4 → msasim-24.12.0}/msasim.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: msasim
-Version: 24.10.4
+Version: 24.12.0
 Summary: A fast MSA simulator
 Home-page: https://github.com/elyawy/Sailfish-backend
 Author: Elya Wygoda

{msasim-24.10.4 → msasim-24.12.0}/setup.py

@@ -9,7 +9,7 @@ from datetime import datetime
 now = datetime.now()
 
 
-__version__ = f"{now.year % 100}.{now.month+1}.4"
+__version__ = f"{now.year % 100}.{now.month+1}.0"
 
 # The main interface is through Pybind11Extension.
 # * You can add cxx_std=11/14/17, and then build_ext can be removed.

{msasim-24.10.4 → msasim-24.12.0}/src/main.cpp

@@ -134,6 +134,7 @@ PYBIND11_MODULE(_Sailfish, m) {
         .def("run_sim", &Simulator::runSimulator)
         .def("init_substitution_sim", &Simulator::initSubstitionSim)
         .def("gen_substitutions", &Simulator::simulateSubstitutions)
+        .def("gen_substitutions_to_dir", &Simulator::simulateAndWriteSubstitutions)
         .def("save_site_rates", &Simulator::setSaveRates)
         .def("get_site_rates", &Simulator::getSiteRates)
         .def("save_all_nodes_sequences", &Simulator::setSaveAllNodes)
@@ -148,9 +149,11 @@ PYBIND11_MODULE(_Sailfish, m) {
         .def("length", &MSA::getMSAlength)
         .def("num_sequences", &MSA::getNumberOfSequences)
         .def("fill_substitutions", &MSA::fillSubstitutions)
+        .def("set_substitutions_folder", &MSA::setSubstitutionsFolder)
         .def("print_msa", &MSA::printFullMsa)
         .def("print_indels", &MSA::printIndels)
         .def("write_msa", &MSA::writeFullMsa)
+        .def("write_msa_from_dir", &MSA::writeMsaFromDir)
         .def("get_msa_string", &MSA::generateMsaString)
         .def("get_msa", &MSA::getMSAVec);