pg-sui 1.0.2.1__py3-none-any.whl → 1.6.8__py3-none-any.whl

test/pg_sui_testing.py

@@ -1,523 +0,0 @@
-#!/usr/bin/env python
-
-# Standard library imports
-import argparse
-import sys
-
-import numpy as np
-import pandas as pd
-import scipy.stats as stats
-
-from sklearn_genetic.space import Continuous, Categorical, Integer
-
-# from pgsui import GenotypeData
-from snpio import GenotypeData
-from impute.estimators import (
-    ImputeNLPCA,
-    ImputeUBP,
-    ImputeRandomForest,
-    ImputeVAE,
-)
-from impute.simple_imputers import ImputePhylo
-
-# from snpio import GenotypeData
-# from impute.estimators import *
-# from impute.simple_imputers import ImputeAlleleFreq, ImputePhylo
-
-# from read_input import GenotypeData
-# from estimators import *
-
-
-def main():
-    """Class instantiations and main package body"""
-
-    args = get_arguments()
-
-    if args.str and args.phylip:
-        sys.exit("Error: Only one file type can be specified")
-
-        # If VCF file is specified.
-    if args.str:
-        if not args.pop_ids and args.popmap is None:
-            raise TypeError("Either --pop_ids or --popmap must be specified\n")
-
-        if args.pop_ids:
-            print("\n--pop_ids was specified as column 2\n")
-        else:
-            print(
-                "\n--pop_ids was not specified; "
-                "using popmap file to get population IDs\n"
-            )
-
-        if args.onerow_perind:
-            print("\nUsing one row per individual...\n")
-        else:
-            print("\nUsing two rows per individual...\n")
-
-        if args.onerow_perind:
-            data = GenotypeData(
-                filename=args.str,
-                filetype="structure1row",
-                popmapfile=args.popmap,
-                guidetree=args.treefile,
-                qmatrix_iqtree=args.iqtree,
-            )
-        else:
-            data = GenotypeData(
-                filename=args.str,
-                filetype="structure2row",
-                popmapfile=args.popmap,
-                guidetree=args.treefile,
-                qmatrix_iqtree=args.iqtree,
-            )
-
-    if args.phylip:
-        if args.pop_ids or args.onerow_perind:
-            print(
-                "\nPhylip file was used with structure arguments; ignoring "
-                "structure file arguments\n"
-            )
-
-        if args.popmap is None:
-            raise TypeError("No popmap file supplied with PHYLIP file\n")
-
-        data = GenotypeData(
-            filename=args.phylip,
-            filetype="phylip",
-            popmapfile=args.popmap,
-            guidetree=args.treefile,
-            qmatrix_iqtree=args.iqtree,
-            siterates_iqtree="pgsui/example_data/trees/test_n10.rate",
-        )
-
-    if args.resume_imputed:
-        pass
-        # data.read_imputed(args.resume_imputed, impute_methods="rf")
-        # data.write_imputed(data.imputed_rf_df, args.prefix)
-
-    else:
-        # For randomizedsearchcv
-        # Number of trees in random forest
-        n_estimators = [
-            int(x) for x in np.linspace(start=100, stop=1000, num=10)
-        ]
-
-        # Number of features to consider at every split
-        max_features = ["sqrt", "log2"]
-
-        # Maximum number of levels in the tree
-        max_depth = [int(x) for x in np.linspace(10, 110, num=11)]
-        max_depth.append(None)
-
-        # Minimmum number of samples required to split a node
-        min_samples_split = [int(x) for x in np.linspace(2, 10, num=5)]
-
-        # Minimum number of samples required at each leaf node
-        min_samples_leaf = [int(x) for x in np.linspace(1, 5, num=5)]
-
-        # Proportion of dataset to use with bootstrapping
-        # max_samples = [x for x in np.linspace(0.5, 1.0, num=6)]
-
-        # # Random Forest gridparams - RandomizedSearchCV
-        # grid_params = {
-        #     "max_features": max_features,
-        #     "max_depth": max_depth,
-        #     "min_samples_split": min_samples_split,
-        #     "min_samples_leaf": min_samples_leaf,
-        # }
-
-        # Random Forest gridparams - Genetic Algorithms
-        # grid_params = {
-        # 	"n_estimators": Integer(100, 500),
-        # 	"max_features": max_features,
-        # 	"max_depth": max_depth,
-        # 	"min_samples_split": min_samples_split,
-        # 	"min_samples_leaf": min_samples_leaf,
-        # 	"max_samples": max_samples
-        # }
-
-        # # Genetic Algorithm grid_params
-        # grid_params = {
-        #     "max_features": Categorical(["sqrt", "log2"]),
-        #     "min_samples_split": Integer(2, 10),
-        #     "min_samples_leaf": Integer(1, 10),
-        #     "max_depth": Integer(2, 110),
-        # }
-
-        # Bayesian Ridge gridparams - RandomizedSearchCV
-        # grid_params = {
-        # 	"alpha_1": stats.loguniform(1e-6, 0.01),
-        # 	"alpha_2": stats.loguniform(1e-6, 0.01),
-        # 	"lambda_1": stats.loguniform(1e-6, 0.01),
-        # 	"lambda_2": stats.loguniform(1e-6, 0.01),
-        # }
-
-        # # Bayesian Ridge gridparams - Genetic algorithm
-        # grid_params = {
-        # 	"alpha_1": Continuous(1e-6, 1e-3, distribution="log-uniform"),
-        # 	"alpha_2": Continuous(1e-6, 1e-3, distribution="log-uniform"),
-        # 	"lambda_1": Continuous(1e-6, 1e-3, distribution="log-uniform"),
-        # 	"lambda_2": Continuous(1e-6, 1e-3, distribution="log-uniform")
-        # }
-
-        # # Random forest imputation with genetic algorithm grid search
-        # rf_imp = ImputeRandomForest(
-        #     data,
-        #     prefix=args.prefix,
-        #     n_estimators=50,
-        #     n_nearest_features=1,
-        #     gridparams=grid_params,
-        #     cv=3,
-        #     grid_iter=40,
-        #     n_jobs=4,
-        #     max_iter=2,
-        #     column_subset=1.0,
-        #     ga=False,
-        #     disable_progressbar=True,
-        #     extratrees=False,
-        #     mutation_probability=0.1,
-        #     progress_update_percent=20,
-        #     chunk_size=1.0,
-        #     initial_strategy="phylogeny",
-        # )
-
-        # # Genetic Algorithm grid search Test
-        # rf_imp2 = ImputeRandomForest(
-        #     data,
-        #     prefix=args.prefix,
-        #     n_estimators=50,
-        #     n_nearest_features=2,
-        #     gridparams=grid_params,
-        #     cv=3,
-        #     grid_iter=40,
-        #     n_jobs=-1,
-        #     max_iter=2,
-        #     column_subset=1.0,
-        #     ga=True,
-        #     disable_progressbar=True,
-        #     extratrees=False,
-        #     chunk_size=1.0,
-        #     initial_strategy="phylogeny",
-        # )
-
-        # rfdata = rf_imp.imputed
-        # print(rfdata.genotypes012_df)
-
-        # rf_data = rf_imp.imputed
-        # print(data.genotypes012_df)
-        # print(rf_data.genotypes012_df)
-
-        # imp_decoded = data.decode_imputed(rf_imp.imputed)
-        # print(imp_decoded)
-
-        # # RandomizedSearchCV Test
-        # rf_imp = ImputeRandomForest(
-        #     data,
-        #     prefix=args.prefix,
-        #     n_estimators=50,
-        #     n_nearest_features=3,
-        #     gridparams=grid_params,
-        #     cv=3,
-        #     grid_iter=40,
-        #     n_jobs=4,
-        #     max_iter=2,
-        #     column_subset=5,
-        #     ga=False,
-        #     disable_progressbar=False,
-        #     extratrees=False,
-        #     progress_update_percent=20,
-        #     chunk_size=0.2,
-        #     initial_strategy="phylogeny",
-        # )
-
-        # lgbm = ImputeLightGBM(
-        #     data,
-        #     prefix=args.prefix,
-        #     cv=3,
-        #     n_jobs=4,
-        #     n_estimators=50,
-        #     disable_progressbar=True,
-        #     chunk_size=0.2,
-        #     validation_only=0.1,
-        #     n_nearest_features=3,
-        #     max_iter=2,
-        #     initial_strategy="populations",
-        # )
-
-        # vae = ImputeVAE(
-        #     genotype_data=data,
-        #     prefix=args.prefix,
-        #     disable_progressbar=True,
-        #     validation_only=None,
-        #     initial_strategy="populations",
-        # )
-
-        # vae_gtdata = vae.imputed
-        # print(vae_gtdata.genotypes012_df)
-
-        # complete_encoded = imputer.train(train_epochs=300, batch_size=256)
-        # print(complete_encoded)
-
-        # rf_imp = ImputeRandomForest(
-        #     data,
-        #     prefix=args.prefix,
-        #     n_estimators=50,
-        #     n_nearest_features=3,
-        #     n_jobs=4,
-        #     max_iter=2,
-        #     disable_progressbar=True,
-        #     extratrees=False,
-        #     max_features="sqrt",
-        #     min_samples_split=5,
-        #     min_samples_leaf=2,
-        #     max_depth=30,
-        #     cv=3,
-        #     validation_only=0.3,
-        #     chunk_size=1.0,
-        #     initial_strategy="populations",
-        # )
-
-        # afpops = ImputeAlleleFreq(
-        #     genotype_data=data,
-        #     by_populations=True,
-        #     prefix=args.prefix,
-        # )
-
-        # print(data.genotypes012_df)
-        # print(afpops.genotypes012_df)
-
-        # br_imp = ImputeBayesianRidge(data, prefix=args.prefix, n_iter=100, gridparams=grid_params, grid_iter=3, cv=3, n_jobs=4, max_iter=5, n_nearest_features=3, column_subset=4, ga=False, disable_progressbar=True, progress_update_percent=20, chunk_size=1.0)
-
-        # aftestpops = ImputeAlleleFreq(
-        #     genotype_data=data, by_populations=True, prefix=args.prefix
-        # )
-
-        # aftestpops_data = aftestpops.imputed
-
-        # print(data.genotypes012_df)
-        # print(aftestpops_data.genotypes012_df)
-
-    # vae = ImputeVAE(
-    #     gt=np.array([[0, 1], [-9, 1], [2, -9]]),
-    #     initial_strategy="most_frequent",
-    #     cv=3,
-    #     validation_only=None,
-    # )
-
-    # vae_data = vae.imputed
-
-    # print(data.genotypes012_df)
-    # print(vae_data.genotypes012_df)
-
-    # For GridSearchCV. Generate parameters to sample from.
-    learning_rate = [float(10) ** x for x in np.arange(-4, 0)]
-    l1_penalty = [float(10) ** x for x in np.arange(-6, -1)]
-    l1_penalty.append(0.0)
-    l2_penalty = [float(10) ** x for x in np.arange(-6, -1)]
-    l2_penalty.append(0.0)
-    hidden_activation = ["elu", "relu"]
-    num_hidden_layers = [1, 2, 3, 4, 5]
-    hidden_layer_sizes = ["sqrt", "midpoint"]
-    n_components = [2, 3]
-    dropout_rate = [round(x, 1) for x in np.arange(0.0, 1.0, 0.1)]
-    batch_size = [16, 32, 48, 64]
-    optimizer = ["adam", "sgd", "adagrad"]
-
-    # grid_params = {
-    #     "learning_rate": Continuous(1e-6, 0.1, distribution="log-uniform"),
-    #     "l2_penalty": Continuous(1e-6, 0.01, distribution="uniform"),
-    #     "n_components": Integer(2, 3),
-    #     # "hidden_activation": Categorical(["elu", "relu"]),
-    # }
-
-    grid_params = {
-        # "learning_rate": learning_rate,
-        # "l1_penalty": l1_penalty,
-        "l2_penalty": l2_penalty,
-        # "hidden_activation": hidden_activation,
-        # "hidden_layer_sizes": hidden_layer_sizes,
-        "n_components": n_components,
-        # "dropout_rate": dropout_rate,
-        # "batch_size": batch_size,
-        # "optimizer": optimizer,
-    }
-
-    ubp = ImputeUBP(
-        data,
-        disable_progressbar=False,
-        cv=3,
-        column_subset=1.0,
-        validation_split=0.0,
-        learning_rate=0.1,
-        num_hidden_layers=1,
-        verbose=1,
-        dropout_rate=0.2,
-        hidden_activation="elu",
-        batch_size=64,
-        l1_penalty=1e-6,
-        l2_penalty=1e-6,
-        gridparams=grid_params,
-        n_jobs=4,
-        grid_iter=5,
-        sim_strategy="nonrandom_weighted",
-        sim_prop_missing=0.4,
-        scoring_metric="precision_recall_macro",
-        gridsearch_method="randomized_gridsearch",
-        early_stop_gen=5,
-        # sample_weights={0: 1.0, 1: 0.0, 2: 1.0},
-        # sample_weights="auto",
-    )
-
-    # ubp = ImputeVAE(
-    #     data,
-    #     # gridparams=grid_params,
-    #     # initial_strategy="populations",
-    #     # disable_progressbar=True,
-    #     # cv=3,
-    #     # column_subset=1.0,
-    #     # validation_size=0.3,
-    #     # learning_rate=0.1,
-    #     # num_hidden_layers=1,
-    #     # verbose=1,
-    #     # gridparams=grid_params,
-    # )
-
-    # nlpca_data = nlpca.imputed
-    # print(nlpca_data.genotypes012_df)
-
-    # print(data.genotypes012_df)
-    # print(nlpca_data.genotypes012_df)
-
-    # ubp = ImputeUBP(
-    #     genotype_data=data,
-    #     test_categorical=np.array([[0, 1], [-9, 1], [2, -9]]),
-    # )
-
-    # ubp = ImputeVAE(
-    #     gt=np.array([[0, 1], [-9, 1], [2, -9]]),
-    #     initial_strategy="most_frequent",
-    # )
-
-    # br_imp = ImputeBayesianRidge(
-    #     data,
-    #     prefix=args.prefix,
-    #     alpha_1=0.0002689638465560243,
-    #     alpha_2=0.0001473822173361299,
-    #     lambda_1=0.0003281735206234651,
-    #     lambda_2=0.00020767920087590963,
-    #     n_iter=100,
-    #     n_nearest_features=3,
-    #     progress_update_percent=20,
-    #     disable_progressbar=True,
-    #     max_iter=2,
-    #     cv=3,
-    #     initial_strategy="group_mode",
-    # )
-
-    # phylo = ImputePhylo(
-    #     genotype_data=data, save_plots=False, disable_progressbar=True
-    # )
-
-    # phylodata = phylo.imputed
-    # print(phylodata.genotypes012_df)
-
-
-def get_arguments():
-    """[Parse command-line arguments. Imported with argparse]
-
-    Returns:
-        [argparse object]: [contains command-line arguments; accessed as method]
-    """
-
-    parser = argparse.ArgumentParser(
-        description="Machine learning missing data imputation and species delimitation",
-        add_help=False,
-    )
-
-    required_args = parser.add_argument_group("Required arguments")
-    filetype_args = parser.add_argument_group(
-        "File type arguments (choose only one)"
-    )
-    structure_args = parser.add_argument_group("Structure file arguments")
-    optional_args = parser.add_argument_group("Optional arguments")
-
-    # File Type arguments
-    filetype_args.add_argument(
-        "-s", "--str", type=str, required=False, help="Input structure file"
-    )
-    filetype_args.add_argument(
-        "-p", "--phylip", type=str, required=False, help="Input phylip file"
-    )
-
-    filetype_args.add_argument(
-        "-t",
-        "--treefile",
-        type=str,
-        required=False,
-        default=None,
-        help="Newick-formatted treefile",
-    )
-
-    filetype_args.add_argument(
-        "-i",
-        "--iqtree",
-        type=str,
-        required=False,
-        help=".iqtree output file containing Rate Matrix Q",
-    )
-
-    # Structure Arguments
-    structure_args.add_argument(
-        "--onerow_perind",
-        default=False,
-        action="store_true",
-        help="Toggles on one row per individual option in structure file",
-    )
-    structure_args.add_argument(
-        "--pop_ids",
-        default=False,
-        required=False,
-        action="store_true",
-        help="Toggles on population ID column (2nd col) in structure file",
-    )
-
-    ## Optional Arguments
-    optional_args.add_argument(
-        "-m",
-        "--popmap",
-        type=str,
-        required=False,
-        default=None,
-        help="Two-column tab-separated population map file: inds\tpops. No header line",
-    )
-    optional_args.add_argument(
-        "--prefix",
-        type=str,
-        required=False,
-        default="output",
-        help="Prefix for output files",
-    )
-
-    optional_args.add_argument(
-        "--resume_imputed",
-        type=str,
-        required=False,
-        help="Read in imputed data from a file instead of doing the imputation",
-    )
-    # Add help menu
-    optional_args.add_argument(
-        "-h", "--help", action="help", help="Displays this help menu"
-    )
-
-    # If no command-line arguments are called then exit and call help menu.
-    if len(sys.argv) == 1:
-        print("\nExiting because no command-line options were called.\n")
-        parser.print_help(sys.stderr)
-        sys.exit(1)
-
-    args = parser.parse_args()
-    return args
-
-
-if __name__ == "__main__":
-    main()