Merge branch 'dev' into footer_calculate-back-next

protzilla/data_analysis/differential_expression.py

@@ -2,10 +2,17 @@
 from .differential_expression_anova import anova
 from .differential_expression_linear_model import linear_model
 from .differential_expression_t_test import t_test
+from .differential_expression_mann_whitney import mann_whitney_test_on_intensity_data, mann_whitney_test_on_ptm_data
+from .differential_expression_kruskal_wallis import kruskal_wallis_test_on_intensity_data, kruskal_wallis_test_on_ptm_data
+
 
 
 # call methods for precommit hook not to delete imports
 def unused():
     anova(**{})
     t_test(**{})
     linear_model(**{})
+    mann_whitney_test_on_ptm_data(**{})
+    mann_whitney_test_on_intensity_data(**{})
+    kruskal_wallis_test_on_intensity_data(**{})
+    kruskal_wallis_test_on_ptm_data(**{})

protzilla/data_analysis/differential_expression_anova.py

@@ -9,7 +9,7 @@
 from .differential_expression_helper import (
     INVALID_PROTEINGROUP_DATA_MSG,
     _map_log_base,
-    apply_multiple_testing_correction,
+    apply_multiple_testing_correction, preprocess_grouping,
 )
 
 
@@ -57,30 +57,7 @@ def anova(
             - a df filtered_proteins, containing the filtered out proteins (due to missing values or identical values),
     """
 
-    assert grouping in metadata_df.columns, f"{grouping} not found in metadata_df"
-    messages = []
-    # Check if the selected groups are present in the metadata_df
-
-    # Select all groups if none or less than two were selected
-    if not selected_groups or isinstance(selected_groups, str):
-        selected_groups = metadata_df[grouping].unique()
-        selected_groups_str = "".join([" " + str(group) for group in selected_groups])
-        messages.append(
-            {
-                "level": logging.INFO,
-                "msg": f"Auto-selected the groups {selected_groups_str} for comparison because none or only one group was selected.",
-            }
-        )
-    elif len(selected_groups) >= 2:
-        for group in selected_groups:
-            if group not in metadata_df[grouping].unique():
-                messages.append(
-                    {
-                        "level": logging.ERROR,
-                        "msg": f"Group {group} not found in metadata_df.",
-                    }
-                )
-                return {"messages": messages}
+    selected_groups, messages = preprocess_grouping(metadata_df, grouping, selected_groups)
 
     # Merge the intensity and metadata dataframes in order to assign to each Sample
     # their corresponding group

protzilla/data_analysis/differential_expression_helper.py

@@ -1,11 +1,14 @@
 import logging
 import math
 
+import numpy as np
+import pandas as pd
+
 from statsmodels.stats.multitest import multipletests
 
 
 def apply_multiple_testing_correction(
-    p_values: list, method: str, alpha: float
+        p_values: list, method: str, alpha: float
 ) -> tuple:
     """
     Applies a multiple testing correction method to a list of p-values
@@ -46,8 +49,96 @@ def _map_log_base(log_base: str) -> int | None:
     return log_base_mapping.get(log_base, None)
 
 
+def preprocess_grouping(
+        metadata_df: pd.DataFrame, grouping: str, selected_groups: list | str
+) -> tuple[list, list[dict]]:
+    """
+    Preprocesses the grouping column in the metadata_df and checks if the selected groups are present.
+    :param metadata_df: the metadata dataframe
+    :param grouping: the column name in the metadata_df that contains the grouping information
+    :param selected_groups: the groups that should be compared
+    :return: a tuple containing the selected groups and a list of messages
+    """
+    assert grouping in metadata_df.columns, f"{grouping} not found in metadata_df"
+    messages = []
+
+    # Check if the selected groups are present in the metadata_df
+    removed_groups = []
+    for group in selected_groups:
+        if group not in metadata_df[grouping].unique():
+            selected_groups.remove(group)
+            removed_groups.append(group)
+    if removed_groups:
+        messages.append(
+            {
+                "level": logging.WARNING,
+                "msg": f"Group{'s' if len(removed_groups) > 1 else ''} "
+                       f"{str(removed_groups)[1:-1]} were not found in metadata_df and thus removed.",
+            }
+        )
+
+    # Select all groups if none or less than two were selected
+    if not selected_groups or isinstance(selected_groups, str) or len(selected_groups) < 2:
+        selected_groups = metadata_df[grouping].unique()
+        selected_groups_str = "".join(["\'" + str(group) + "\', " for group in selected_groups])[0:-2]
+        messages.append(
+            {
+                "level": logging.WARNING,
+                "msg": f"Auto-selected the groups {selected_groups_str} for comparison, "
+                       f"because none or only one {'valid ' if removed_groups else ''}group was selected.",
+            }
+        )
+
+    return selected_groups, messages
+
+
+def calculate_log2_fold_change(
+        group1_data: pd.Series, group2_data: pd.Series, log_base: int
+) -> float:
+    return (
+        np.log2(
+            np.power(log_base, group2_data).mean()
+            / np.power(log_base, group1_data).mean()
+        )
+        if log_base
+        else np.log2(group2_data.mean() / group1_data.mean())
+    )
+
+
+def merge_differential_expression_and_significant_df(
+        intensity_df: pd.DataFrame, diff_exp_df: pd.DataFrame, sig_df: pd.DataFrame
+) -> tuple[pd.DataFrame, pd.DataFrame]:
+    differentially_expressed_proteins_df = pd.merge(intensity_df, diff_exp_df, on="Protein ID", how="left")
+    differentially_expressed_proteins_df = differentially_expressed_proteins_df.loc[
+        differentially_expressed_proteins_df["Protein ID"].isin(diff_exp_df["Protein ID"])
+    ]
+    significant_proteins_df = pd.merge(intensity_df, sig_df, on="Protein ID", how="left")
+    significant_proteins_df = significant_proteins_df.loc[
+        significant_proteins_df["Protein ID"].isin(sig_df["Protein ID"])
+    ]
+
+    return differentially_expressed_proteins_df, significant_proteins_df
+
+
+def normalize_ptm_df(ptm_df: pd.DataFrame) -> pd.DataFrame:
+    """
+    Normalizes the PTM data frame by dividing the PTM values by the amount of peptides.
+    :param ptm_df: the PTM data frame
+    :return: the normalized PTM data frame
+    """
+    ptm_df_without_sample = ptm_df.drop("Sample", axis=1)
+
+    normalized_ptm_df = ptm_df_without_sample.div(ptm_df["Total Amount of Peptides"], axis=0)
+
+    normalized_ptm_df = normalized_ptm_df.drop("Total Amount of Peptides", axis=1)
+    normalized_ptm_df.insert(0, "Sample", ptm_df["Sample"])
+
+    return normalized_ptm_df
+
+
 INVALID_PROTEINGROUP_DATA_MSG = {
     "level": logging.WARNING,
-    "msg": "Due do missing or identical values, the p-values for some protein groups could not be calculated. These groups were omitted from the analysis. "
-    "To prevent this, please add filtering and imputation steps to your workflow before running the analysis.",
+    "msg": "Due to missing or identical values, the p-values for some protein groups could not be calculated. "
+           "These groups were omitted from the analysis. "
+           "To prevent this, please add filtering and imputation steps to your workflow before running the analysis.",
 }