color selection for enrichment analysis bar plot

protzilla/data_integration/di_plots.py

@@ -15,9 +15,8 @@ def GO_enrichment_bar_plot(
     top_terms,
     cutoff,
     value,
-    gene_sets=[],
+    gene_sets={},
     title="",
-    colors=PLOT_COLOR_SEQUENCE,
     figsize=None,
 ):
     """
@@ -73,8 +72,6 @@ def GO_enrichment_bar_plot(
     if not gene_sets:
         msg = "Please select at least one category to plot."
         return dict(messages=[dict(level=logging.ERROR, msg=msg)])
-    if not isinstance(gene_sets, list):
-        gene_sets = [gene_sets]
     if value not in ["fdr", "p-value"]:
         msg = "Invalid value. Value must be either 'fdr' or 'p-value'."
         return dict(messages=[dict(level=logging.ERROR, msg=msg)])
@@ -108,9 +105,8 @@ def GO_enrichment_bar_plot(
     elif value == "p-value":
         column = "P-value" if restring_input else "Adjusted P-value"
 
+    colors = gene_sets.values()
 
-    if colors == "" or colors is None or len(colors) == 0:
-        colors = PLOT_COLOR_SEQUENCE
     size_y = top_terms * 0.5 * len(gene_sets)
     try:
         ax = gseapy.barplot(

protzilla/methods/data_integration.py

@@ -234,7 +234,6 @@ class PlotGOEnrichmentBarPlot(PlotStep):
         "top_terms",
         "cutoff",
         "title",
-        "colors",
         "figsize",
     ]
     # TODO: input figsize optional?

tests/protzilla/data_integration/test_plots_data_integration.py

@@ -25,7 +25,7 @@ def test_enrichment_bar_plot_restring(show_figures, helpers):
         top_terms=10,
         cutoff=0.05,
         value="fdr",
-        gene_sets=["KEGG", "Process"],
+        gene_sets={"KEGG" : "#E2A46D", "Process" : "#4A536A"},
     )
     if show_figures:
         helpers.open_graph_from_base64(bar_base64[0])
@@ -35,7 +35,7 @@ def test_enrichment_bar_plot_restring(show_figures, helpers):
         top_terms=10,
         cutoff=0.05,
         value="p_value",
-        gene_sets=["KEGG", "Process"],
+        gene_sets={"KEGG" : "#E2A46D", "Process" : "#4A536A"},
     )
     if show_figures:
         helpers.open_graph_from_base64(bar_base64[0])
@@ -50,7 +50,7 @@ def test_enrichment_bar_plot(show_figures, helpers, data_folder_tests):
         top_terms=10,
         cutoff=0.05,
         value="p_value",
-        gene_sets=["Reactome_2013"],
+        gene_sets={"Reactome_2013" : "#E2A46D"},
     )
     if show_figures:
         helpers.open_graph_from_base64(bar_base64[0])
@@ -65,7 +65,7 @@ def test_enrichment_bar_plot_wrong_value(data_folder_tests):
         top_terms=10,
         cutoff=0.05,
         value="fdr",
-        gene_sets=["Reactome_2013"],
+        gene_sets={"Reactome_2013" : "#E2A46D"},
     )
     assert "messages" in current_out
     assert any(("FDR is not available" in message["msg"]) for message in current_out["messages"])
@@ -78,7 +78,7 @@ def test_enrichment_bar_plot_empty_df():
         top_terms=10,
         cutoff=0.05,
         value="p_value",
-        gene_sets=["Reactome_2013"],
+        gene_sets={"Reactome_2013" : "#E2A46D"},
     )
     assert "messages" in current_out
     assert any(("No data to plot" in message["msg"]) for message in current_out["messages"])
@@ -102,7 +102,7 @@ def test_enrichment_bar_plot_wrong_df():
         top_terms=10,
         cutoff=0.05,
         value="p_value",
-        gene_sets=["KEGG"],
+        gene_sets={"KEGG" : "#E2A46D"},
     )
     assert "messages" in current_out
     assert any(("Please choose an enrichment result dataframe" in message["msg"]) for message in current_out["messages"])
@@ -115,7 +115,7 @@ def test_enrichment_bar_plot_cutoff(data_folder_tests):
         top_terms=10,
         cutoff=0,
         value="fdr",
-        gene_sets=["KEGG", "Process"],
+        gene_sets={"KEGG" : "#E2A46D", "Process" : "#4A536A"},
     )
 
     assert "messages" in current_out
@@ -129,7 +129,7 @@ def test_enrichment_bar_plot_cutoff(data_folder_tests):
         top_terms=10,
         cutoff=0,
         value="p-value",
-        gene_sets=["Reactome_2013"],
+        gene_sets={"Reactome_2013" : "#E2A46D"},
     )
     assert "messages" in current_out
     assert any(("No data to plot when applying cutoff" in message["msg"]) for message in current_out["messages"])

ui/runs/forms/custom_fields.py

@@ -13,6 +13,7 @@
 from django.forms.widgets import CheckboxInput, SelectMultiple
 from django.utils.html import format_html
 from django.utils.safestring import SafeText, mark_safe
+from django.template.loader import render_to_string
 
 # Custom widgets
 
@@ -48,7 +49,7 @@ def __init__(self, choices: Enum | list, initial=None, *args, **kwargs):
             super().__init__(choices=choices, initial=initial, *args, **kwargs)
         else:
             super().__init__(
-                choices=[(el.value, el.value) for el in choices],
+                choices=[(choice.value, choice.value) for choice in choices],
                 initial=initial,
                 *args,
                 **kwargs,
@@ -76,7 +77,7 @@ def __init__(self, choices: Enum | list, initial=None, *args, **kwargs):
             super().__init__(choices=choices, initial=initial, *args, **kwargs)
         else:
             super().__init__(
-                choices=[(el.value, el.value) for el in choices],
+                choices=[(choice.value, choice.value) for choice in choices],
                 initial=initial,
                 *args,
                 **kwargs,
@@ -85,7 +86,70 @@ def __init__(self, choices: Enum | list, initial=None, *args, **kwargs):
         self.widget.attrs.update({"class": "form-select mb-2"})
 
     def clean(self, value: list[str] | None):
-        return [el for el in value if el != "hidden"] if value else None
+        return [element for element in value if element != "hidden"] if value else None
+
+class CustomCheckboxMultipleChoiceField(MultipleChoiceField):
+    def __init__(self, choices: Enum | list, colors: Enum | list, initial=None, *args, **kwargs):
+        if isinstance(choices, list):
+            super().__init__(choices=choices, initial=initial, *args, **kwargs)
+        else:
+            super().__init__(
+                choices=[(choice.value, choice.value) for choice in choices],
+                initial=initial,
+                *args,
+                **kwargs,
+            )
+        self.widget = CustomCheckboxSelectMultipleWidget()
+        self.widget.colors = colors
+        self.widget.attrs.update({"class": "form-select mb-2"})
+
+
+    def clean(self, value: list[str] | None):
+        if not value:
+            return None
+
+        gen_sets = []
+        colors = {}
+        result = {}
+
+        for element in value:
+            if element.startswith("color_"):
+                _,gen_set,color = element.split('_', 2)
+                colors[gen_set] = color
+            else:
+                gen_sets.append(element)
+
+        for gen_set in sorted(gen_sets):
+            if gen_set in colors:
+                result[gen_set] = colors[gen_set]
+
+        return result
+
+
+class CustomCheckboxSelectMultipleWidget(SelectMultiple):
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.colors = []
+
+    def render(self, name, value, attrs=None, renderer=None) -> SafeText:
+        if isinstance(value, dict):
+            reformat_value = []
+            for gen_set, color in value.items():
+                reformat_value.append(gen_set)
+                reformat_value.append(f"color_{gen_set}_{color}")
+            value = reformat_value
+
+        return mark_safe(
+            render_to_string(
+                "runs/field_component_color_selection.html",
+                context={
+                    "name": name,
+                    "choices": self.choices,
+                    "colors": self.colors,
+                    "values": value,
+                    }
+            )
+        )
 
 
 class CustomFileField(FileField):

ui/runs/forms/data_integration.py

@@ -22,6 +22,7 @@
     CustomFloatField,
     CustomMultipleChoiceField,
     CustomNumberField,
+    CustomCheckboxMultipleChoiceField,
 )
 
 PROTEIN_DF = "protein_df"
@@ -215,7 +216,7 @@ def fill_form(self, run: Run) -> None:
             self.toggle_visibility("gene_sets_enrichr", True)
             self.fields["gene_sets_enrichr"].choices = fill_helper.to_choices(
                 gseapy.get_library_name()
-            )  # TODO check whether we need to pass the organism name here
+            )  #  check whether we need to pass the organism name here
         else:
             self.toggle_visibility("gene_sets_path", True)
 
@@ -324,7 +325,7 @@ class EnrichmentAnalysisWithGSEAForm(MethodForm):
         choices=GeneSetsField,
         label="How do you want to provide the gene sets? (reselect to show dynamic fields)",
         initial="Choose from Enrichr options"
-        # Todo: Dynamic parameters
+        # : Dynamic parameters
     )
     gene_sets_path = CustomFileField(
         label="Upload gene sets with uppercase gene symbols (any of the following file "
@@ -334,15 +335,15 @@ class EnrichmentAnalysisWithGSEAForm(MethodForm):
         "SetName2: [Gene2, Gene3, ...]})",
         initial=None,
     )
-    # Todo: gene_sets_enrichr dynamic filling
+    # : gene_sets_enrichr dynamic filling
     gene_sets_enrichr = CustomChoiceField(choices=[], label="Gene sets")
     grouping = CustomChoiceField(
         choices=[],
         label="Grouping from metadata",
         initial=None
-        # Todo: Dynamic parameters
+        # : Dynamic parameters
     )
-    # Todo: add dynamic filling to group1, group2
+    # : add dynamic filling to group1, group2
     group1 = CustomChoiceField(choices=[], label="Group1", initial=None)
 
     group2 = CustomChoiceField(choices=[], label="Group2", initial=None)
@@ -402,7 +403,7 @@ def fill_form(self, run: Run) -> None:
             self.toggle_visibility("gene_sets_enrichr", True)
             self.fields["gene_sets_enrichr"].choices = fill_helper.to_choices(
                 gseapy.get_library_name()
-            )  # TODO check whether we need to pass the organism name here
+            )  #  check whether we need to pass the organism name here
         else:
             self.toggle_visibility("gene_sets_path", True)
 
@@ -446,20 +447,20 @@ def fill_form(self, run: Run) -> None:
 
 
 class EnrichmentAnalysisWithPrerankedGSEAForm(MethodForm):
-    # Todo: protein_df
-    # Todo: ranking_column
+    # : protein_df
+    # : ranking_column
     ranking_direction = CustomChoiceField(
         choices=RankingDirectionField,
         label="Sort the ranking column (ascending - smaller values are better, "
         "descending - larger values are better)",
         initial=RankingDirectionField.ascending,
     )
-    # Todo: gene_mapping
+    # : gene_mapping
     gene_sets_field = CustomChoiceField(
         choices=GeneSetsField,
         label="How do you want to provide the gene sets? (reselect to show dynamic fields)",
         initial=GeneSetsField.choose_from_enrichr_options
-        # Todo: Dynamic parameters
+        # : Dynamic parameters
     )
     gene_sets_path = CustomFileField(
         label="Upload gene sets with uppercase gene symbols (any of the following file "
@@ -469,7 +470,7 @@ class EnrichmentAnalysisWithPrerankedGSEAForm(MethodForm):
         "SetName2: [Gene2, Gene3, ...]})",
         initial=None,
     )
-    # Todo: gene_sets_enrichr
+    # : gene_sets_enrichr
     min_size = CustomNumberField(
         label="Minimum number of genes from gene set also in data", initial=15
     )
@@ -513,7 +514,7 @@ class DatabaseIntegrationByGeneMappingForm(MethodForm):
     )
     dataframe = CustomChoiceField(
         choices=[], label="Step to use"
-    )  # TODO this looks and sounds very generic, be more specific, maybe it needs diffexp step
+    )  #  this looks and sounds very generic, be more specific, maybe it needs diffexp step
 
     def fill_form(self, run: Run) -> None:
         self.fields["database_names"].choices = fill_helper.to_choices(
@@ -525,8 +526,8 @@ def fill_form(self, run: Run) -> None:
 
 
 class DatabaseIntegrationByUniprotForm(MethodForm):
-    # Todo: uniprot
-    # Todo: Add dynamic fill for database name and fields
+    # : uniprot
+    # : Add dynamic fill for database name and fields
     database_name = CustomChoiceField(
         choices=[],
         label="Uniprot databases (offline)",
@@ -535,11 +536,12 @@ class DatabaseIntegrationByUniprotForm(MethodForm):
 
 
 class PlotGOEnrichmentBarPlotForm(MethodForm):
-    # TODO: input:df fill dynamic with fill_forms
+    # : input:df fill dynamic with fill_forms
     input_df_step_instance = CustomChoiceField(
         choices=[], label="Choose dataframe to be plotted"
     )
-    gene_sets = CustomMultipleChoiceField(choices=[], label="Sets to be plotted")
+    # TODO: after the color naming has been optimised in all filese, the underlying line can be updated: (color, color) for color in PLOT_COLOR_SEQUENCE
+    gene_sets = CustomCheckboxMultipleChoiceField(choices=[], colors=[(v, k[4:]) for k, v, in list(mcolors.TABLEAU_COLORS.items())], label="Sets to be plotted")
     value = CustomChoiceField(
         choices=GOEnrichmentBarPlotValue,
         label="Value (bars will be plotted as -log10(value)), fdr only for GO analysis with STRING, p_value is adjusted if available",
@@ -561,15 +563,7 @@ class PlotGOEnrichmentBarPlotForm(MethodForm):
     )
     title = CustomCharField(label="Title of the plot (optional)", required=False)
 
-    colors = CustomMultipleChoiceField(
-        choices=[], label="Colors for the plot (optional)"
-    )  # TODO this should  not have to be set in fill_form
-
     def fill_form(self, run: Run) -> None:
-        self.fields["colors"].choices = [
-            (v, k) for k, v, in mcolors.CSS4_COLORS.items()
-        ]
-
         self.fields["input_df_step_instance"].choices = fill_helper.get_choices(
             run, "enrichment_df"
         )

ui/runs/templates/runs/field_component_color_selection.html

@@ -0,0 +1,27 @@
+<div class="checkbox-container border p-3 rounded">
+    {% for option_value, option_label in choices %}
+    <div class="d-flex" style="align-items: center">
+        <div class="checkbox-item px-2 w-50">
+            <input type="checkbox" id="{{ name }}_{{ option_label}}_checkbox" name="{{ name }}" value="{{ option_value }}" {% if option_value in values %}checked{% endif %}>
+            <label for="{{ name }}_{{ option_label}}_checkbox">{{ option_label }}</label>
+        </div>
+        <div class="ms-auto px-2 w-50">
+            <select class="form-select" name="{{ name }}">
+                {% for color, color_label in colors %}
+                    {% with color_value="color_"|add:option_value|add:"_"|add:color %}
+                        <option value={{ color_value }}
+                        {% if values %}
+                            {% if color_value in values %}selected{% endif %}
+                        {% else %}
+                            {% if forloop.parentloop.counter0 == forloop.counter0 %}selected{% endif %}
+                        {% endif %} 
+                        >
+                            {{ color_label }}
+                        </option>
+                    {% endwith %}
+                {% endfor %}
+            </select>
+        </div>
+    </div>
+    {% endfor %}
+</div>