Merge pull request #489 from Proteobench/mean_median_plot

Change median or mean quant error viz depending on user's choice

proteobench/datapoint/quant_datapoint.py

@@ -185,7 +185,9 @@ def get_metrics(df: pd.DataFrame, min_nr_observed: int = 1) -> Dict[int, Dict[st
         nr_prec = len(df_slice)
 
         # Calculate the median absolute epsilon (insensitive to outliers)
-        median_abs_epsilon = df_slice["epsilon"].abs().mean()
+        median_abs_epsilon = df_slice["epsilon"].abs().median()
+        # Calculate the mean absolute epsilon (sensitive to outliers)
+        mean_abs_epsilon = df_slice["epsilon"].abs().mean()
 
         # Calculate the variance of epsilon (sensitive to outliers)
         variance_epsilon = df_slice["epsilon"].var()
@@ -199,6 +201,7 @@ def get_metrics(df: pd.DataFrame, min_nr_observed: int = 1) -> Dict[int, Dict[st
         return {
             min_nr_observed: {
                 "median_abs_epsilon": median_abs_epsilon,
+                "mean_abs_epsilon": mean_abs_epsilon,
                 "variance_epsilon": variance_epsilon,
                 "nr_prec": nr_prec,
                 "CV_median": cv_median,

proteobench/plotting/plot_quant.py

@@ -71,6 +71,7 @@ def plot_fold_change_histogram(result_df: pd.DataFrame, species_ratio: Dict[str,
     @staticmethod
     def plot_metric(
         benchmark_metrics_df: pd.DataFrame,
+        metric: str = "median",
         software_colors: Dict[str, str] = {
             "MaxQuant": "#377eb8",
             "AlphaPept": "#4daf4a",
@@ -107,6 +108,7 @@ def plot_metric(
         all_median_abs_epsilon = [
             v2["median_abs_epsilon"] for v in benchmark_metrics_df["results"] for v2 in v.values()
         ]
+        all_mean_abs_epsilon = [v2["mean_abs_epsilon"] for v in benchmark_metrics_df["results"] for v2 in v.values()]
         all_nr_prec = [v2["nr_prec"] for v in benchmark_metrics_df["results"] for v2 in v.values()]
 
         # Add hover text with detailed information for each data point
@@ -123,6 +125,7 @@ def plot_metric(
                         datapoint_text + f"Comment (private submission): {benchmark_metrics_df.comments[idx]}"
                     )
             else:
+                # TODO: Determine parameters based on module
                 datapoint_text = (
                     f"ProteoBench ID: {benchmark_metrics_df.id[idx]}<br>"
                     + f"Software tool: {benchmark_metrics_df.software_name[idx]} {benchmark_metrics_df.software_version[idx]}<br>"
@@ -162,15 +165,27 @@ def plot_metric(
         benchmark_metrics_df["hover_text"] = hover_texts
         benchmark_metrics_df["scatter_size"] = scatter_size
 
+        if metric == "median":
+            layout_xaxis_range = [
+                min(all_median_abs_epsilon) - min(all_median_abs_epsilon) * 0.05,
+                max(all_median_abs_epsilon) + min(all_median_abs_epsilon) * 0.05,
+            ]
+            layout_xaxis_title = (
+                "Median absolute difference between measured and expected log2-transformed fold change."
+            )
+        elif metric == "mean":
+            layout_xaxis_range = [
+                min(all_mean_abs_epsilon) - min(all_mean_abs_epsilon) * 0.05,
+                max(all_mean_abs_epsilon) + min(all_mean_abs_epsilon) * 0.05,
+            ]
+            layout_xaxis_title = "Mean absolute difference between measured and expected log2-transformed fold change."
+
         fig = go.Figure(
             layout_yaxis_range=[
                 min(all_nr_prec) - min(max(all_nr_prec) * 0.05, 2000),
                 max(all_nr_prec) + min(max(all_nr_prec) * 0.05, 2000),
             ],
-            layout_xaxis_range=[
-                min(all_median_abs_epsilon) - min(all_median_abs_epsilon) * 0.05,
-                max(all_median_abs_epsilon) + min(all_median_abs_epsilon) * 0.05,
-            ],
+            layout_xaxis_range=layout_xaxis_range,
         )
 
         # Get all unique color-software combinations (necessary for highlighting)
@@ -205,7 +220,7 @@ def plot_metric(
             width=700,
             height=700,
             xaxis=dict(
-                title="Mean absolute difference between measured and expected log2-transformed fold change",
+                title=layout_xaxis_title,
                 gridcolor="white",
                 gridwidth=2,
                 linecolor="black",

webinterface/pages/base_pages/quant.py

@@ -129,9 +129,17 @@ def display_submitted_results(self) -> None:
         self.initialize_submitted_data_points()
         data_points_filtered = self.filter_data_submitted_slider()
 
+        metric = st.radio(
+            "Select metric to plot",
+            options=["Median", "Mean"],
+            help="Toggle between median and mean absolute difference metrics.",
+            key="placeholder2",  # TODO: add to variables
+        )
+
         try:
             fig_metric = PlotDataPoint.plot_metric(
                 data_points_filtered,
+                metric=metric,
                 label=st.session_state[st.session_state[self.variables_quant.selectbox_id_submitted_uuid]],
             )
             st.plotly_chart(fig_metric, use_container_width=True)
@@ -155,9 +163,18 @@ def display_existing_results(self) -> None:
         self.initialize_main_data_points()
         data_points_filtered = self.filter_data_main_slider()
 
+        metric = st.radio(
+            "Select metric to plot",
+            options=["Median", "Mean"],
+            help="Toggle between median and mean absolute difference metrics.",
+        )
+
         try:
             fig_metric = PlotDataPoint.plot_metric(
-                data_points_filtered, label=st.session_state[st.session_state[self.variables_quant.selectbox_id_uuid]]
+                data_points_filtered,
+                label=st.session_state[st.session_state[self.variables_quant.selectbox_id_uuid]],
+                metric=metric,
+                key="placeholder",  # TODO: add to variables
             )
             st.plotly_chart(fig_metric, use_container_width=True)
         except Exception as e: