msasim 24.10.4__tar.gz → 24.13.0__tar.gz

{msasim-24.10.4 → msasim-24.13.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: msasim
-Version: 24.10.4
+Version: 24.13.0
 Summary: A fast MSA simulator
 Home-page: https://github.com/elyawy/Sailfish-backend
 Author: Elya Wygoda

{msasim-24.10.4 → msasim-24.13.0}/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)
     
@@ -465,15 +462,16 @@ class Simulator:
     def set_replacement_model(
             self,
             model: _Sailfish.modelCode,
+            amino_model_file: pathlib.Path = None,
             model_parameters: List = None,
             gamma_parameters_alpha : float = 1.0,
-            gamma_parameters_catergories: int = 1,
+            gamma_parameters_categories: int = 1,
             invariant_sites_proportion: float = 0.0
         ) -> None:
         if not model:
             raise ValueError(f"please provide a substitution model from the the following list: {_Sailfish.modelCode}")
-        if int(gamma_parameters_catergories) != gamma_parameters_catergories:
-            raise ValueError(f"gamma_parameters_catergories has to be a positive int value: received value of {gamma_parameters_catergories}")
+        if int(gamma_parameters_categories) != gamma_parameters_categories:
+            raise ValueError(f"gamma_parameters_catergories has to be a positive int value: received value of {gamma_parameters_categories}")
         self._model_factory = _Sailfish.modelFactory(self._simProtocol._get_Sailfish_tree())
 
         self._model_factory.set_alphabet(self._alphabet)
@@ -481,6 +479,8 @@ class Simulator:
             if model_parameters:
                 raise ValueError(f"no model parameters are used in protein, recevied value of: {model_parameters}")
             self._model_factory.set_replacement_model(model)
+            if model == MODEL_CODES.CUSTOM and amino_model_file:
+                self._model_factory.set_amino_replacement_model_file(str(amino_model_file))
         else:
             if model == MODEL_CODES.NUCJC and model_parameters:
                 raise ValueError(f"no model parameters in JC model, recevied value of: {model_parameters}")
@@ -492,7 +492,7 @@ class Simulator:
             else:
                 self._model_factory.set_model_parameters(model_parameters)
 
-        self._model_factory.set_gamma_parameters(gamma_parameters_alpha, gamma_parameters_catergories)
+        self._model_factory.set_gamma_parameters(gamma_parameters_alpha, gamma_parameters_categories)
         self._model_factory.set_invariant_sites_proportion(invariant_sites_proportion)
         self._simulator.init_substitution_sim(self._model_factory)
 
@@ -537,6 +537,26 @@ 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-24.13.0}/msasim.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: msasim
-Version: 24.10.4
+Version: 24.13.0
 Summary: A fast MSA simulator
 Home-page: https://github.com/elyawy/Sailfish-backend
 Author: Elya Wygoda

{msasim-24.10.4 → msasim-24.13.0}/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-24.13.0}/src/main.cpp

@@ -113,6 +113,7 @@ PYBIND11_MODULE(_Sailfish, m) {
         .value("EX_EHO_EXP_EXT", modelCode::EX_EHO_EXP_EXT)
         .value("EX_EHO_EXP_HEL", modelCode::EX_EHO_EXP_HEL)
         .value("EX_EHO_EXP_OTH", modelCode::EX_EHO_EXP_OTH)
+        .value("CUSTOM", modelCode::CUSTOM)
         .export_values();
 
 
@@ -121,6 +122,7 @@ PYBIND11_MODULE(_Sailfish, m) {
         .def(py::init<tree*>())
         .def("set_alphabet", &modelFactory::setAlphabet)
         .def("set_replacement_model" , &modelFactory::setReplacementModel)
+        .def("set_amino_replacement_model_file" , &modelFactory::setCustomAAModelFile)
         .def("set_model_parameters" , &modelFactory::setModelParameters)
         .def("set_gamma_parameters" , &modelFactory::setGammaParameters)
         .def("set_invariant_sites_proportion", &modelFactory::setInvariantSitesProportion)
@@ -134,6 +136,7 @@ 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 +151,11 @@ 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);
 

{msasim-24.10.4 → msasim-24.13.0}/src/modelFactory.cpp

@@ -30,7 +30,7 @@ void modelFactory::setReplacementModel(modelCode model) {
     _state = factoryState::PARAMETERS;
     if (_alphabet == alphabetCode::AMINOACID) _state = factoryState::GAMMA;
     if (_model == modelCode::AAJC || _model == modelCode::NUCJC) _state = factoryState::GAMMA;
-
+    if (_model == modelCode::CUSTOM) _state = factoryState::MODEL_FILE;
 }
 
 void modelFactory::setModelParameters(std::vector<MDOUBLE> params) {
@@ -63,6 +63,15 @@ void modelFactory::setModelParameters(std::vector<MDOUBLE> params) {
     _state = factoryState::GAMMA;
 }
 
+void modelFactory::setCustomAAModelFile(const std::string &fileName) {
+    if (_state!=factoryState::MODEL_FILE) {
+        std::cout << "Please set the model to 'CUSTOM' before proceeding.\n";
+        return;
+    }
+    _modelFilePath = fileName;
+    _state = factoryState::GAMMA;
+}
+
 void modelFactory::setGammaParameters(MDOUBLE alpha, size_t numCategories) {
     if (_state!=factoryState::GAMMA) {
         std::cout << "Please specify a model and its correct parameters before proceeding.\n";
@@ -184,6 +193,21 @@ std::shared_ptr<stochasticProcess> modelFactory::getStochasticProcess() {
         case modelCode::EX_EHO_EXP_OTH:
             repModel = std::make_unique<pupAll>(datMatrixHolder::EX_EHO_EXP_OTH);
             break;
+        case modelCode::CUSTOM:
+            std::ifstream in(_modelFilePath);
+            if (!in.is_open()) throw std::runtime_error("Could not open file");
+            std::stringstream contents;
+            char buffer;
+            while (in.get(buffer)) {
+                if (buffer == '\"' || buffer == '\n') continue;
+                contents << buffer;
+            }
+            in.close();
+            const std::string &tmpstr = contents.str();
+            const char* cstr = tmpstr.c_str();
+            datMatrixString aminoFileString(cstr);
+            repModel = std::make_unique<pupAll>(aminoFileString);
+            break;
     }
 
     std::unique_ptr<pijAccelerator> pij;