adding accuracy modelling

pydamage/accuracy_model.py

@@ -1,15 +1,15 @@
-from pypmml import Model
 import pkg_resources
 import pandas as pd
 import numpy as np
+import pickle
 
 
 def load_model():
     """Returns the pmml model"""
     # This is a stream,like object. If you want the actual info, call
     # stream.read()
-    stream = pkg_resources.resource_stream(__name__, "models/accuracy_model.xml")
-    return Model.load(stream)
+    stream = pkg_resources.resource_stream(__name__, "models/glm_accuracy_model.pickle")
+    return pickle.load(stream)
 
 
 def prepare_data(pd_df):
@@ -70,20 +70,15 @@ def prepare_data(pd_df):
     ]
     simu_cov = pd.cut(pd_df["coverage"], coverage_bins)
     simu_cov.cat.rename_categories(coverage_bins_labels, inplace=True)
-    pd_df["simu_cov"] = simu_cov
+    pd_df["simuCov"] = simu_cov
 
     simu_contig_length = pd.cut(pd_df["reflen"], reflen_bins)
     simu_contig_length.cat.rename_categories(reflen_bins_labels, inplace=True)
 
-    pd_df["simu_contig_length"] = simu_contig_length
+    pd_df["simuContigLength"] = simu_contig_length
     pd_df = pd_df[
-        [
-            "simu_cov",
-            "simu_contig_length",
-            "damage_model_pmax",
-            "gc_content",
-        ]
-    ].rename(columns={"damage_model_pmax": "damage"})
+        ["simuCov", "simuContigLength", "damage_model_pmax", "gc_content"]
+    ].rename(columns={"damage_model_pmax": "damage", "gc_content": "GCcontent"})
 
     return pd_df
 
@@ -95,4 +90,4 @@ def fit_model(df, model):
         df (pandas DataFrame): prepared pydamage results
         model (pypmml model): GLM accuracy model
     """
-    return model.predict(df)
+    return model.predict(df).to_frame(name="pred_accuracy")

pydamage/main.py

@@ -109,12 +109,13 @@ def analyze(
         with multiprocessing.Pool(proc) as p:
             list(tqdm(p.imap(plot_partial, filt_res), total=len(filt_res)))
 
-    df = utils.pandas_processing(res_dict=filt_res, outdir=outdir)
+    df_pydamage = utils.pandas_processing(res_dict=filt_res)
 
     acc_model = load_model()
-    print(acc_model)
-    prep_df = prepare_data(df)
-    print(prep_df)
-    print(fit_model(prep_df, acc_model))
+    prep_df_glm = prepare_data(df_pydamage)
+    df_glm = fit_model(prep_df_glm, acc_model)
 
+    df = df_pydamage.merge(df_glm, left_index=True, right_index=True)
+
+    utils.df_to_csv(df, outdir)
     return df

pydamage/utils.py

@@ -46,12 +46,11 @@ def makedir(dirpath, confirm=True, force=False):
     os.makedirs(dirpath)
 
 
-def pandas_processing(res_dict, outdir):
+def pandas_processing(res_dict):
     """Performs Pandas processing of Pydamage results
 
     Args:
         res_dict (dict): Result dictionary of Vuong's closeness test
-        outdir (str): Path to output directory
     """
     df = pd.DataFrame(res_dict)
     if len(res_dict) == 0:
@@ -101,10 +100,19 @@ def pandas_processing(res_dict, outdir):
     df.set_index("reference", inplace=True)
     df.dropna(axis=1, how="all", inplace=True)
     df = df.round(3)
-    df.to_csv(f"{outdir}/pydamage_results.csv")
     return df
 
 
+def df_to_csv(df, outdir):
+    """Write Pydamage results to disk
+
+    Args:
+        df(pandas DataFrame): Pydamage results DataFrame
+        outdir (str): Path to output directory
+    """
+    df.to_csv(f"{outdir}/pydamage_results.csv")
+
+
 def sort_dict_by_keys(adict):
     """Sort dictonary by keys
 

setup.py

@@ -36,11 +36,10 @@ def get_version(rel_path):
         "statsmodels",
         "matplotlib",
         "tqdm",
-        "pypmml",
         "biopython",
     ],
     packages=find_packages(include=["pydamage"]),
     entry_points={"console_scripts": ["pydamage = pydamage.cli:cli"]},
     include_package_data=True,
-    package_data={"": ["models/accuracy_model.xml"]},
+    package_data={"": ["models/glm_accuracy_model.pickle"]},
 )