Refactor analysis i

alphastats/gui/gui.py

@@ -18,6 +18,8 @@ def run():
     os.system(
         f"python -m streamlit run AlphaPeptStats.py --global.developmentMode=false {extra_args}"
     )
+
+    # TODO why are we starting the app a second time here?
     _this_file = os.path.abspath(__file__)
     _this_directory = os.path.dirname(_this_file)
 

alphastats/gui/pages/04_Analysis.py

@@ -42,7 +42,7 @@
 show_df = False
 analysis_result = None
 
-c1, *_ = st.columns([1, 1, 1])
+c1, c2 = st.columns([0.33, 0.67])
 with c1:
     method = st.selectbox(
         "Analysis",
@@ -66,21 +66,21 @@
         )
         show_df = analysis_result is not None
 
+with c2:
+    # --- SHOW PLOT -------------------------------------------------------
+    if show_plot:
+        display_plot(method, analysis_result)
 
-# --- SHOW PLOT -------------------------------------------------------
-if show_plot:
-    display_plot(method, analysis_result)
+    # --- SHOW STATISTICAL ANALYSIS -------------------------------------------------------
+    elif show_df:
+        display_df(analysis_result)
 
-# --- SHOW STATISTICAL ANALYSIS -------------------------------------------------------
-elif show_df:
-    display_df(analysis_result)
+        csv = convert_df(analysis_result)
 
-    csv = convert_df(analysis_result)
-
-    # TODO do we want to save statistical analysis to results page as well?
-    st.download_button(
-        "Download as .csv", csv, method + ".csv", "text/csv", key="download-csv"
-    )
+        # TODO do we want to save statistical analysis to results page as well?
+        st.download_button(
+            "Download as .csv", csv, method + ".csv", "text/csv", key="download-csv"
+        )
 
 
 @st.fragment

alphastats/gui/utils/analysis.py

@@ -0,0 +1,155 @@
+"""Module providing frontend widgets for gathering parameters and mapping them to the actual analysis."""
+
+from abc import ABC, abstractmethod
+from collections import defaultdict
+
+import streamlit as st
+
+from alphastats.DataSet import DataSet
+from alphastats.keys import Cols
+from alphastats.plots.VolcanoPlot import VolcanoPlot
+
+
+class Analysis(ABC):
+    """Abstract class for analysis widgets."""
+
+    def __init__(self, dataset):
+        self._dataset: DataSet = dataset
+        self._parameters = defaultdict(lambda: None)
+
+    @abstractmethod
+    def show_widget(self):
+        """Show the widget and gather parameters."""
+        pass
+
+    @abstractmethod
+    def do_analysis(self):
+        """Perform the analysis.
+
+        Returns a tuple(figure, analysis_object, parameters) where figure is the plot,
+        analysis_object is the underlying object, parameters is a dictionary of the parameters used.
+        """
+        pass
+
+
+class GroupCompareAnalysis(Analysis, ABC):
+    """Abstract class for group comparison analysis widgets."""
+
+    def show_widget(self):
+        """Gather parameters to compare two group."""
+
+        metadata = self._dataset.metadata
+
+        default_option = "<select>"
+        custom_group_option = "Custom group from samples .."
+
+        grouping_variable = st.selectbox(
+            "Grouping variable",
+            options=[default_option]
+            + metadata.columns.to_list()
+            + [custom_group_option],
+        )
+
+        column = None
+        if grouping_variable == default_option:
+            st.stop()  # TODO: using stop here is not really great
+        elif grouping_variable != custom_group_option:
+            unique_values = metadata[grouping_variable].unique().tolist()
+
+            column = grouping_variable
+            group1 = st.selectbox("Group 1", options=unique_values)
+            group2 = st.selectbox("Group 2", options=list(reversed(unique_values)))
+
+        else:
+            group1 = st.multiselect(
+                "Group 1 samples:",
+                options=metadata[Cols.SAMPLE].to_list(),
+            )
+
+            group2 = st.multiselect(
+                "Group 2 samples:",
+                options=list(reversed(metadata[Cols.SAMPLE].to_list())),
+            )
+
+            intersection_list = list(set(group1).intersection(set(group2)))
+            if len(intersection_list) > 0:
+                st.warning(
+                    "Group 1 and Group 2 contain same samples: "
+                    + str(intersection_list)
+                )
+
+        if group1 == group2:
+            st.error(
+                "Group 1 and Group 2 can not be the same. Please select different groups."
+            )
+            st.stop()
+
+        self._parameters.update({"group1": group1, "group2": group2})
+        if column is not None:
+            self._parameters["column"] = column
+
+
+class VolcanoPlotAnalysis(GroupCompareAnalysis):
+    """Widget for Volcano Plot analysis."""
+
+    def show_widget(self):
+        """Show the widget and gather parameters."""
+        super().show_widget()
+
+        parameters = {}
+        method = st.selectbox(
+            "Differential Analysis using:",
+            options=["ttest", "anova", "wald", "sam", "paired-ttest", "welch-ttest"],
+        )
+        parameters["method"] = method
+
+        parameters["labels"] = st.checkbox("Add labels", value=True)
+
+        parameters["draw_line"] = st.checkbox("Draw lines", value=True)
+
+        parameters["alpha"] = st.number_input(
+            label="alpha", min_value=0.001, max_value=0.050, value=0.050
+        )
+
+        parameters["min_fc"] = st.select_slider(
+            "Foldchange cutoff", range(0, 3), value=1
+        )
+
+        if method == "sam":
+            parameters["perm"] = st.number_input(
+                label="Number of Permutations", min_value=1, max_value=1000, value=10
+            )
+            parameters["fdr"] = st.number_input(
+                label="FDR cut off", min_value=0.005, max_value=0.1, value=0.050
+            )
+
+        self._parameters.update(parameters)
+
+    def do_analysis(self):
+        """Draw Volcano Plot using the VolcanoPlot class.
+
+        Returns a tuple(figure, analysis_object, parameters) where figure is the plot,
+        analysis_object is the underlying object, parameters is a dictionary of the parameters used.
+        """
+        # TODO currently there's no other way to obtain both the plot and the underlying data
+        #  Should be refactored such that the interface provided by DateSet.plot_volcano() is used
+        #  One option could be to always return the whole analysis object.
+
+        volcano_plot = VolcanoPlot(
+            mat=self._dataset.mat,
+            rawinput=self._dataset.rawinput,
+            metadata=self._dataset.metadata,
+            preprocessing_info=self._dataset.preprocessing_info,
+            group1=self._parameters["group1"],
+            group2=self._parameters["group2"],
+            column=self._parameters["column"],
+            method=self._parameters["method"],
+            labels=self._parameters["labels"],
+            min_fc=self._parameters["min_fc"],
+            alpha=self._parameters["alpha"],
+            draw_line=self._parameters["draw_line"],
+            perm=self._parameters["perm"],
+            fdr=self._parameters["fdr"],
+            color_list=self._parameters["color_list"],
+        )
+        return volcano_plot.plot, volcano_plot, self._parameters

alphastats/gui/utils/analysis_helper.py

@@ -4,9 +4,9 @@
 import pandas as pd
 import streamlit as st
 
+from alphastats.gui.utils.analysis import VolcanoPlotAnalysis
 from alphastats.gui.utils.ui_helper import StateKeys, convert_df
 from alphastats.keys import Cols
-from alphastats.plots.VolcanoPlot import VolcanoPlot
 
 
 def display_figure(plot):
@@ -118,62 +118,6 @@ def st_general(method_dict):
     return chosen_parameter_dict
 
 
-def gui_volcano_plot() -> Tuple[Optional[Any], Optional[Any], Optional[Dict]]:
-    """Draw Volcano Plot using the VolcanoPlot class.
-
-    Returns a tuple(figure, analysis_object, parameters) where figure is the plot,
-    analysis_object is the underlying object, parameters is a dictionary of the parameters used.
-    """
-    parameters = helper_compare_two_groups()
-    method = st.selectbox(
-        "Differential Analysis using:",
-        options=["ttest", "anova", "wald", "sam", "paired-ttest", "welch-ttest"],
-    )
-    parameters.update({"method": method})
-
-    labels = st.checkbox("Add label")
-    parameters.update({"labels": labels})
-
-    draw_line = st.checkbox("Draw line")
-    parameters.update({"draw_line": draw_line})
-
-    alpha = st.number_input(
-        label="alpha", min_value=0.001, max_value=0.050, value=0.050
-    )
-    parameters.update({"alpha": alpha})
-
-    min_fc = st.select_slider("Foldchange cutoff", range(0, 3), value=1)
-    parameters.update({"min_fc": min_fc})
-
-    if method == "sam":
-        perm = st.number_input(
-            label="Number of Permutations", min_value=1, max_value=1000, value=10
-        )
-        fdr = st.number_input(
-            label="FDR cut off", min_value=0.005, max_value=0.1, value=0.050
-        )
-        parameters.update({"perm": perm, "fdr": fdr})
-
-    submitted = st.button("Run analysis ..")
-
-    if submitted:
-        dataset = st.session_state[StateKeys.DATASET]
-
-        # TODO currently there's no other way to obtain both the plot and the underlying data
-        #  Should be refactored such that the interface provided by DateSet.plot_volcano() is used
-        #  One option could be to alyways return the whole analysis object.
-        volcano_plot = VolcanoPlot(
-            mat=dataset.mat,
-            rawinput=dataset.rawinput,
-            metadata=dataset.metadata,
-            preprocessing_info=dataset.preprocessing_info,
-            **parameters,
-        )
-        return volcano_plot.plot, volcano_plot, parameters
-
-    return None, None, None
-
-
 def do_analysis(
     method: str, options_dict: Dict[str, Any]
 ) -> Tuple[Optional[Any], Optional[Any], Dict[str, Any]]:
@@ -185,11 +129,15 @@ def do_analysis(
     Currently, analysis_object is only not-None for Volcano Plot.
     # TODO unify the API of all analysis methods
     """
-
     method_dict = options_dict.get(method)
 
     if method == "Volcano Plot":
-        return gui_volcano_plot()
+        analysis = VolcanoPlotAnalysis(st.session_state[StateKeys.DATASET])
+        analysis.show_widget()
+
+        if st.button("Run analysis .."):
+            return analysis.do_analysis()
+        return None, None, {}
 
     elif method == "t-SNE Plot":
         parameters = st_tsne_options(method_dict)
@@ -235,6 +183,7 @@ def helper_plot_dimensionality_reduction(method_dict):
     return chosen_parameter_dict
 
 
+# TODO this can be deleted after all analysis adapted the new Pattern (cf. analysis.py:Analysis())
 def helper_compare_two_groups():
     """
     Helper function to compare two groups for example

alphastats/gui/utils/options.py

@@ -71,10 +71,7 @@ def get_plotting_options(state):
             },
             "function": dataset.plot_tsne,
         },
-        "Volcano Plot": {
-            "between_two_groups": True,  # TODO: between_two_groups is never used anywhere
-            "function": dataset.plot_volcano,
-        },
+        "Volcano Plot": {},
         "Clustermap": {"function": dataset.plot_clustermap},
         # "Dendrogram": {"function": state[StateKeys.DATASET].plot_dendrogram},  # TODO why commented?
     }

alphastats/llm/enrichment_analysis.py

@@ -4,10 +4,6 @@
 import requests
 from gprofiler import GProfiler
 
-# TODO needed?
-# from Bio import Entrez
-# Entrez.email = "lebedev_mikhail@outlook.com"  # Always provide your email address when using NCBI services.
-
 
 def _get_functional_annotation_string(identifier, species_id="9606") -> pd.DataFrame:
     """

alphastats/statistics/MultiCovaAnalysis.py

@@ -18,7 +18,7 @@ def __init__(
         subset: dict = None,
         plot: bool = False,
     ):
-        self.dataset = dataset  # TODO pass only .mat, .metadata and .sample
+        self.dataset = dataset  # TODO pass only .mat, .metadata
         self.covariates = covariates
         self.n_permutations = n_permutations
         self.fdr = fdr

requirements.txt

@@ -1,5 +1,4 @@
 anndata==0.9.1
-biopython==1.83
 click==8.0.1
 combat==0.3.3
 data_cache>=0.1.6