Idhandling I

alphastats/dataset/dataset.py

@@ -1,3 +1,4 @@
+from collections import defaultdict
 from typing import Dict, List, Optional, Tuple, Union
 
 import pandas as pd
@@ -101,23 +102,11 @@ def __init__(
         self.mat: pd.DataFrame = mat
         self.metadata: pd.DataFrame = metadata
         self.preprocessing_info: Dict = preprocessing_info
-
-        self._gene_name_to_protein_id_map = (
-            {
-                k: v
-                for k, v in dict(
-                    zip(
-                        self.rawinput[Cols.GENE_NAMES].tolist(),
-                        self.rawinput[Cols.INDEX].tolist(),
-                    )
-                ).items()
-                if isinstance(k, str)  # avoid having NaN as key
-            }
-            if Cols.GENE_NAMES in self.rawinput.columns
-            else {}
-        )
-        # TODO This is not necessarily unique, and should ideally raise an error in some of our test-data sets that
-        #  contain isoform ids. E.g. TPM1 occurs 5 times in testfiles/maxquant/proteinGroups.txt with different base Protein IDs.
+        (
+            self._gene_to_features_map,
+            self._protein_to_features_map,
+            self._feature_to_repr_map,
+        ) = self._create_id_dicts()
 
         print("DataSet has been created.")
 
@@ -161,6 +150,37 @@ def _check_loader(loader):
                 "Invalid index_column: consider reloading your data with: AlphaPeptLoader, MaxQuantLoader, DIANNLoader, FragPipeLoader, SpectronautLoader"
             )
 
+    def _create_id_dicts(self, sep: str = ";") -> Tuple[dict, dict, dict]:
+        """Create mapprings from gene, protein to feature and from feature to repr."""
+
+        features = self.mat.columns.to_list()
+        gene_to_features_map = defaultdict(lambda: [])
+        protein_to_features_map = defaultdict(lambda: [])
+        feature_to_repr_map = defaultdict(lambda x: x)
+
+        for proteins, feature in self.rawinput[[Cols.INDEX, Cols.INDEX]].itertuples(
+            index=False
+        ):
+            if feature not in features:
+                continue
+            # TODO: Shorten list if too many ids e.g. to id1;...(19) if 20 ids are present
+            feature_to_repr_map[feature] = "ids:" + proteins
+            for protein in proteins.split(sep):
+                protein_to_features_map[protein].append(feature)
+
+        if Cols.GENE_NAMES in self.rawinput.columns:
+            for genes, feature in self.rawinput[
+                [Cols.GENE_NAMES, Cols.INDEX]
+            ].itertuples(index=False):
+                if feature not in features:
+                    continue
+                if isinstance(genes, str):
+                    for gene in genes.split(sep):
+                        gene_to_features_map[gene].append(feature)
+                    feature_to_repr_map[feature] = genes
+
+        return gene_to_features_map, protein_to_features_map, feature_to_repr_map
+
     def _get_preprocess(self) -> Preprocess:
         """Return instance of the Preprocess object."""
         return Preprocess(
@@ -199,6 +219,11 @@ def preprocess(
                 **kwargs,
             )
         )
+        (
+            self._gene_to_features_map,
+            self._protein_to_features_map,
+            self._feature_to_repr_map,
+        ) = self._create_id_dicts()
 
     def reset_preprocessing(self):
         """Reset all preprocessing steps"""
@@ -208,6 +233,11 @@ def reset_preprocessing(self):
             self.metadata,
             self.preprocessing_info,
         ) = self._get_init_dataset()
+        (
+            self._gene_to_features_map,
+            self._protein_to_features_map,
+            self._feature_to_repr_map,
+        ) = self._create_id_dicts()
 
     def batch_correction(self, batch: str) -> None:
         """A wrapper for Preprocess.batch_correction(), see documentation there."""
@@ -434,25 +464,22 @@ def plot_volcano(
 
         return volcano_plot.plot
 
-    def _get_protein_id_for_gene_name(
+    def _get_features_for_gene_name(
         self,
         gene_name: str,
-    ) -> str:
-        """Get protein id from gene id. If gene id is not present, return gene id, as we might already have a gene id.
-        'VCL;HEL114' -> 'P18206;A0A024QZN4;V9HWK2;B3KXA2;Q5JQ13;B4DKC9;B4DTM7;A0A096LPE1'
+    ) -> list:
+        """Get protein groups from gene id. If gene id is not present, return gene id, as we might already have a gene id.
+        'HEL114' -> ['P18206;A0A024QZN4;V9HWK2;B3KXA2;Q5JQ13;B4DKC9;B4DTM7;A0A096LPE1']
 
         Args:
             gene_name (str): Gene name
 
         Returns:
-            str: Protein id or gene name if not present in the mapping.
+            list: Protein group ids or gene name if not present in the mapping.
         """
-        if gene_name in self._gene_name_to_protein_id_map:
-            return self._gene_name_to_protein_id_map[gene_name]
+        if gene_name in self._gene_to_features_map:
+            return self._gene_to_features_map[gene_name]
 
-        for gene, protein_id in self._gene_name_to_protein_id_map.items():
-            if gene_name in gene.split(";"):
-                return protein_id
         return gene_name
 
     def plot_intensity(
@@ -492,7 +519,7 @@ def plot_intensity(
         if gene_name is None and protein_id is not None:
             pass
         elif gene_name is not None and protein_id is None:
-            protein_id = self._get_protein_id_for_gene_name(gene_name)
+            protein_id = self._get_features_for_gene_name(gene_name)
         else:
             raise ValueError(
                 "Either protein_id or gene_name must be provided, but not both."

alphastats/gui/pages/05_Analysis.py

@@ -94,7 +94,7 @@ def show_start_llm_button(analysis_method: str) -> None:
         st.session_state[StateKeys.LLM_INPUT] = (analysis_object, parameters)
 
         st.toast("LLM analysis created!", icon="✅")
-        st.page_link("pages/05_LLM.py", label="=> Go to LLM page..")
+        st.page_link("pages/06_LLM.py", label="=> Go to LLM page..")
 
 
 if analysis_result is not None:

alphastats/loader/generic_loader.py

@@ -11,6 +11,7 @@ def __init__(
         file: Union[str, pd.DataFrame],
         intensity_column: list,
         index_column: str,
+        gene_names_column: str = None,
         sep: str = None,
     ):
         """Generic Loader for you proteomics data
@@ -33,7 +34,8 @@ def __init__(
         self.confidence_column = None
         self.software = "Generic"
         self.evidence_df = None
-        self.gene_names_column = None
+        if gene_names_column in self.rawinput.columns.to_list():
+            self.gene_names_column = gene_names_column
         self.ptm_df = None
         self._add_contamination_column()
         self._check_if_columns_are_present()

alphastats/loader/maxquant_loader.py

@@ -1,3 +1,4 @@
+import re
 from typing import Union
 
 import numpy as np
@@ -45,6 +46,23 @@ def __init__(
         self._set_filter_columns_to_true_false()
         self._read_all_column_names_as_string()
 
+        intensity_columns = [
+            col
+            for col in self.rawinput.columns
+            if intensity_column.replace("[sample]", "") in col
+        ]
+        if len(self.rawinput.dropna(subset=intensity_columns, how="all")) != len(
+            self.rawinput
+        ):
+            valid_id = re.compile(
+                "[A-Z]"
+            )  # Assuming that all valid protein ids would contain at least one letter.
+            self.rawinput = self.rawinput[
+                self.rawinput[self.index_column].apply(
+                    lambda x: bool(valid_id.match(x))
+                )
+            ]
+
         if gene_names_column in self.rawinput.columns.to_list():
             self.gene_names_column = gene_names_column
 

alphastats/plots/intensity_plot.py

@@ -9,6 +9,7 @@
 import scipy
 
 from alphastats.dataset.keys import Cols
+from alphastats.dataset.preprocessing import PreprocessingStateKeys
 from alphastats.plots.plot_utils import PlotlyObject, PlotUtils
 
 plotly.io.templates["alphastats_colors"] = plotly.graph_objects.layout.Template(
@@ -42,7 +43,7 @@ def __init__(
         preprocessing_info: Dict,
         protein_id,
         group,
-        subgroups,
+        subgroups=None,
         method,
         add_significance,
         log_scale,
@@ -52,12 +53,15 @@ def __init__(
         self.intensity_column = intensity_column
         self.preprocessing_info = preprocessing_info
 
-        self.protein_id = protein_id
+        self.protein_id = [protein_id] if isinstance(protein_id, str) else protein_id
         self.group = group
         self.subgroups = subgroups
         self.method = method
         self.add_significance = add_significance
         self.log_scale = log_scale
+        self.y_axis = self.intensity_column.replace("[sample]", "").strip()
+        if self.preprocessing_info[PreprocessingStateKeys.LOG2_TRANSFORMED]:
+            self.y_axis = "log2(" + self.yaxis + ")"
 
         self.prepared_df = None
         self._prepare_data()
@@ -137,58 +141,58 @@ def _add_significance(plot):
     def _prepare_data(self):
         # TODO use difflib to find similar ProteinId if ProteinGroup is not present
         df = (
-            self.mat[[self.protein_id]]
-            .reset_index()
-            .rename(columns={"index": Cols.SAMPLE})
-        )
+            self.mat[self.protein_id].melt(
+                ignore_index=False,
+                value_name=self.y_axis,
+                var_name=Cols.INDEX,
+            )
+        ).dropna()
+        df = df.reset_index().rename(columns={"index": Cols.SAMPLE})
         df = df.merge(self.metadata, how="inner", on=[Cols.SAMPLE])
 
         if self.subgroups is not None:
             df = df[df[self.group].isin(self.subgroups)]
 
-        self.y_label = (
-            self.protein_id + " - " + self.intensity_column.replace("[sample]", "")
-        )
         self.prepared_df = df
 
     def _plot(self):
         if self.method == "violin":
             fig = px.violin(
                 self.prepared_df,
-                y=self.protein_id,
+                y=self.y_axis,
                 x=self.group,
+                facet_col=Cols.INDEX,
                 color=self.group,
-                labels={self.protein_id: self.y_label},
                 template="simple_white+alphastats_colors",
             )
 
         elif self.method == "box":
             fig = px.box(
                 self.prepared_df,
-                y=self.protein_id,
+                y=self.y_axis,
                 x=self.group,
+                facet_col=Cols.INDEX,
                 color=self.group,
-                labels={self.protein_id: self.y_label},
                 template="simple_white+alphastats_colors",
             )
 
         elif self.method == "scatter":
             fig = px.scatter(
                 self.prepared_df,
-                y=self.protein_id,
+                y=self.y_axis,
                 x=self.group,
+                facet_col=Cols.INDEX,
                 color=self.group,
-                labels={self.protein_id: self.y_label},
                 template="simple_white+alphastats_colors",
             )
 
         elif self.method == "all":
             fig = px.violin(
                 self.prepared_df,
-                y=self.protein_id,
+                y=self.y_axis,
                 x=self.group,
+                facet_col=Cols.INDEX,
                 color=self.group,
-                labels={self.protein_id: self.y_label},
                 box=True,
                 points="all",
                 template="simple_white+alphastats_colors",

tests/gui/test_02_import_data.py

@@ -106,5 +106,5 @@ def test_page_02_loads_maxquant_testfiles(
 
     dataset = at.session_state[StateKeys.DATASET]
     assert dataset._intensity_column == "LFQ intensity [sample]"
-    assert dataset.rawmat.shape == (312, 2611)
+    assert dataset.rawmat.shape == (312, 2249)
     assert dataset.software == "MaxQuant"