Merge pull request #18 from broadinstitute/add-drug-counts

Add drug counts to data page data availability graph

frontend/packages/portal-frontend/src/dAPI.ts

@@ -657,6 +657,7 @@ export class DepmapApi {
       // mapped to color category integers. The integer maps to Heatmap.tsx's
       // color scale.
       data_type_url_mapping: boolSummary.data_type_url_mapping,
+      drug_count_mapping: boolSummary.drug_count_mapping,
       values: dataAvailVals,
       data_types: boolSummary.data_types,
     };

frontend/packages/portal-frontend/src/dataPage/components/DataAvailabilityPlot.tsx

@@ -97,6 +97,22 @@ const DataAvailabilityPlot = ({
 
     return graphSectionUrlMapping;
   };
+
+  const getDrugCountMapping = (categoryKey: string) => {
+    const graphSectionDrugCountMapping: {
+      [key: string]: number | undefined;
+    } = {};
+    dataValuesByDataTypeCategory[categoryKey].forEach((category: any) => {
+      const dataTypeString = getDataPageDataTypeString(category.dataType);
+
+      const count =
+        currentReleaseDataAvil.drug_count_mapping[category.dataType];
+
+      graphSectionDrugCountMapping[dataTypeString] = count;
+    });
+
+    return graphSectionDrugCountMapping;
+  };
   return (
     <div>
       {dataValuesByDataTypeCategory && (
@@ -112,6 +128,7 @@ const DataAvailabilityPlot = ({
                       (category: any) => category.dataType
                     )}
                     dataTypeUrlMapping={getDataTypeUrlMapping(categoryKey)}
+                    drugCountMapping={getDrugCountMapping(categoryKey)}
                     dataTypeGroupName={categoryKey}
                     isCurrentRelease={isCurrentRelease}
                   />
@@ -131,7 +148,7 @@ const DataAvailabilityPlot = ({
                   BAR_THICKNESS *
                   dataValuesByDataTypeCategory[categoryKey].length
                 }
-                customWidth={400}
+                customWidth={350}
                 customColorScale={COLOR_SCALE}
                 margin={{
                   l: 0,

frontend/packages/portal-frontend/src/dataPage/components/DataPage.tsx

@@ -33,6 +33,7 @@ export const DataPage = ({
   const [allDataAvail, setAllDataAvail] = useState<DataAvailability>({
     all_depmap_ids: [],
     data_type_url_mapping: {},
+    drug_count_mapping: {},
     data_types: [],
     values: [],
   });
@@ -42,6 +43,7 @@ export const DataPage = ({
   ] = useState<DataAvailability>({
     all_depmap_ids: [],
     data_type_url_mapping: {},
+    drug_count_mapping: {},
     data_types: [],
     values: [],
   });
@@ -65,18 +67,22 @@ export const DataPage = ({
       const currentDataValues: number[][] = [];
       const currentDataTypes: string[] = [];
       const currentDataTypeUrlMapping: { [key: string]: string } = {};
+      const currentDataTypeDrugCountMapping: { [key: string]: number } = {};
       dataAvail.data_types.forEach((data_type: string, index: number) => {
         if (currentReleaseDatasets.includes(data_type)) {
           currentDataTypes.push(data_type);
           currentDataValues.push(dataAvail.values[index]);
           currentDataTypeUrlMapping[data_type] =
             dataAvail.data_type_url_mapping[data_type];
+          currentDataTypeDrugCountMapping[data_type] =
+            dataAvail.drug_count_mapping[data_type];
         }
       });
 
       setCurrentReleaseDataAvail({
         all_depmap_ids: dataAvail.all_depmap_ids,
         data_type_url_mapping: currentDataTypeUrlMapping,
+        drug_count_mapping: currentDataTypeDrugCountMapping,
         data_types: currentDataTypes,
         values: currentDataValues,
       });

frontend/packages/portal-frontend/src/dataPage/components/DataPageDatatypeSelector.tsx

@@ -7,17 +7,31 @@ interface Props {
   datatypes: string[];
   dataTypeGroupName: string;
   dataTypeUrlMapping: { [key: string]: string | undefined };
+  drugCountMapping: { [data_type: string]: number | undefined };
   isCurrentRelease: boolean;
 }
 
 const DataPageDatatypeSelector = ({
   datatypes,
   dataTypeGroupName,
   dataTypeUrlMapping,
+  drugCountMapping,
   isCurrentRelease,
 }: Props) => {
   const display: any = [];
 
+  const getDrugCountLabel = (dataType: string) => {
+    if (
+      Object.keys(drugCountMapping).includes(dataType) &&
+      drugCountMapping[dataType] !== null &&
+      drugCountMapping[dataType] !== undefined
+    ) {
+      return `(${drugCountMapping[dataType]} drugs)`;
+    }
+
+    return "";
+  };
+
   // If the data type is in the current release, add an asterisk.
   const getDataTypeLabel = (
     dataType: string,
@@ -47,10 +61,15 @@ const DataPageDatatypeSelector = ({
             target="_blank"
             rel="noreferrer"
           >
-            {getDataTypeLabel(dataTypeDisplayName, inCurrentRelease)}
+            {`${getDataTypeLabel(dataTypeDisplayName, inCurrentRelease)}`}
+            <br />
+            {getDrugCountLabel(dataTypeDisplayName)}
           </a>
         ) : (
-          <p>{getDataTypeLabel(dataTypeDisplayName, inCurrentRelease)}</p>
+          <p>
+            {getDataTypeLabel(dataTypeDisplayName, inCurrentRelease)} <br />
+            {getDrugCountLabel(dataTypeDisplayName)}
+          </p>
         )}
       </div>
     );

frontend/packages/portal-frontend/src/dataPage/components/utils.ts

@@ -1,4 +1,4 @@
-export const BAR_THICKNESS = 42;
+export const BAR_THICKNESS = 43;
 
 export const currentReleaseDatasets = [
   "Sequencing_WES_Broad",

frontend/packages/portal-frontend/src/dataPage/models/types.ts

@@ -21,6 +21,7 @@ export const COLOR_SCALE = [
 export interface DataAvailSummary {
   all_depmap_ids: [number, string][];
   data_type_url_mapping: { [data_type: string]: string };
+  drug_count_mapping: { [data_type: string]: number };
   data_types: string[];
   values: boolean[][];
 }
@@ -38,6 +39,7 @@ export interface DataSummary {
 export interface DataAvailability {
   all_depmap_ids: [number, string][];
   data_type_url_mapping: { [data_type: string]: string };
+  drug_count_mapping: { [data_type: string]: number };
   values: number[][];
   data_types: string[];
 }

frontend/packages/portal-frontend/src/dataPage/styles/DataAvailabilityPlot.scss

@@ -1,30 +1,22 @@
 .dataAvailabilityPlotContainer {
-  width: 290px;
-  min-width: 290px;
+  width: 275px;
+  min-width: 275px;
   margin-top: 2px;
   border-bottom: 1px solid black;
   padding-bottom: 0px;
   margin-right: 8px;
 
-  @media (max-width: 910px) {
-    width: 200px;
-    min-width: 200px;
-  }
   .datatypeSelector {
-    margin-left: 5px;
-    font-size: 16px;
-    min-width: 101px;
-
-    @media (max-width: 910px) {
-      font-size: 14px;
-    }
+    margin-left: 4px;
+    font-size: 14px;
+    min-width: 150px;
 
     .groupContainer {
       display: grid;
       grid-template-columns: repeat(2, 1fr);
       .groupLabel {
         grid-column: 1;
-        margin-right: 20px;
+        margin-right: 10px;
         align-self: normal;
         font-family: roboto;
         font-size: 16px;

frontend/packages/portal-frontend/src/plot/components/Heatmap.tsx

@@ -107,7 +107,7 @@ function Heatmap({
           height,
 
           margin,
-          width: 400,
+          width: customWidth,
         }
       : {
           title: "",

portal-backend/depmap/data_page/api.py

@@ -4,6 +4,7 @@
 from depmap.download.utils import get_download_url
 from depmap.enums import BiomarkerEnum, DependencyEnum
 from flask_restplus import Namespace, Resource
+from depmap.data_access import interface as data_access
 from flask import current_app
 import pandas as pd
 
@@ -39,6 +40,33 @@
 ]
 
 
+def _get_drug_count_mapping(data_types: List[str]):
+    def _get_drug_count(dataset_name: str):
+        dataset = DependencyDataset.get_dataset_by_name(dataset_name)
+
+        if not dataset:
+            return None
+
+        return len(data_access.get_dataset_feature_ids_by_label(dataset_name))
+
+    drug_counts_by_data_type = {
+        "Drug_CTD_Broad": _get_drug_count(DependencyEnum.CTRP_AUC.name),
+        "Drug_Repurposing_Broad": _get_drug_count(
+            DependencyEnum.Rep_all_single_pt.name
+        ),
+        "Drug_GDSC_Sanger": _get_drug_count(DependencyEnum.GDSC2_AUC.name),
+        "Drug_OncRef_Broad": _get_drug_count(DependencyEnum.Prism_oncology_AUC.name),
+    }
+
+    current_env_mapping = {
+        data_type_name: drug_count
+        for data_type_name, drug_count in drug_counts_by_data_type.items()
+        if data_type_name in data_types
+    }
+
+    return current_env_mapping
+
+
 def _get_data_type_url_mapping(data_types: List[str]):
     def _get_dataset_url(dataset_name, isDependencyDataset=True) -> Union[str, None]:
         data_page_endpoint = "data_page.view_data_page"
@@ -118,9 +146,12 @@ def _get_formatted_all_data_avail_df(overall_summary: pd.DataFrame) -> pd.DataFr
 def _format_data_availability_summary_dict(summary_df: pd.DataFrame):
     data_types_by_url = _get_data_type_url_mapping(summary_df.index.values.tolist())
 
+    drug_count_mapping = _get_drug_count_mapping(summary_df.index.values.tolist())
+
     summary = {
         "values": [row.values.tolist() for _, row in summary_df.iterrows()],
         "data_type_url_mapping": data_types_by_url,
+        "drug_count_mapping": drug_count_mapping,
         # For keeping track of data_type order
         "data_types": summary_df.index.values.tolist(),
     }
@@ -145,4 +176,5 @@ def get(self):
         all_data_df = _get_all_data_avail_df()
         formatted_df = _get_formatted_all_data_avail_df(all_data_df)
         all_data_dict = _format_data_availability_summary_dict(formatted_df)
+
         return all_data_dict

portal-backend/tests/depmap/data_page/test_all_data_avail_partial.csv

@@ -1,2 +1,5 @@
 ModelID,Drug_CTD_Broad,Drug_Repurposing_Broad,CRISPR_Achilles_Broad
-ACH-000019,True,False,True
+ACH-000019,False,False,True
+ACH-000020,True,False,False
+ACH-000021,True,False,False
+ACH-000022,True,False,False

portal-backend/tests/depmap/data_page/test_api.py

@@ -6,8 +6,12 @@
     DependencyDatasetFactory,
     DepmapModelFactory,
     GeneFactory,
+    CompoundFactory,
+    CompoundExperimentFactory,
     MatrixFactory,
 )
+import numpy as np
+from tests.utilities import interactive_test_utils
 
 
 def test_get_data_availability(populated_db):
@@ -19,6 +23,7 @@ def test_get_data_availability(populated_db):
         assert list(data_availablity.keys()) == [
             "values",
             "data_type_url_mapping",
+            "drug_count_mapping",
             "data_types",
             "all_depmap_ids",
         ]
@@ -65,7 +70,23 @@ def test_get_data_availability_not_all_data_types_present(
     )
     dataset = DependencyDatasetFactory(matrix=matrix, name=dataset_name, priority=1,)
 
+    drug_dataset_name = DependencyDataset.DependencyEnum.CTRP_AUC
+    cell_lines = [
+        DepmapModelFactory(model_id="ACH-000020"),
+        DepmapModelFactory(model_id="ACH-000021"),
+        DepmapModelFactory(model_id="ACH-000022"),
+    ]
+    compounds = [CompoundFactory() for _ in range(3)]
+    compound_experiments = [CompoundExperimentFactory(compound=c) for c in compounds]
+    matrix = MatrixFactory(
+        data=[[2, 3, 4], [0, 2, 1], [np.NaN, np.NaN, np.NaN]],
+        cell_lines=cell_lines,
+        entities=compound_experiments,
+        using_depmap_model_table=True,
+    )
+    DependencyDatasetFactory(matrix=matrix, name=drug_dataset_name, priority=1)
     empty_db_mock_downloads.session.flush()
+    interactive_test_utils.reload_interactive_config()
 
     def mock_get_all_data_avail_df():
         with open(
@@ -77,22 +98,36 @@ def mock_get_all_data_avail_df():
     monkeypatch.setattr(
         data_page_api, "_get_all_data_avail_df", mock_get_all_data_avail_df
     )
+
     with empty_db_mock_downloads.app.test_client() as c:
         r = c.get(
             url_for("api.data_page_data_availability"), content_type="application/json",
         )
         data_availability = r.json
         assert data_availability == {
-            "values": [[True], [True], [False]],
+            "values": [
+                [True, False, False, False],
+                [False, True, True, True],
+                [False, False, False, False],
+            ],
             "data_type_url_mapping": {
                 "CRISPR_Achilles_Broad": "/data_page/?release=test+name+version&file=test+file+name+2",
-                "Drug_CTD_Broad": None,
+                "Drug_CTD_Broad": "/data_page/?release=test+name+version&file=test+file+name+2",
+                "Drug_Repurposing_Broad": None,
+            },
+            "drug_count_mapping": {
+                "Drug_CTD_Broad": 3,
                 "Drug_Repurposing_Broad": None,
             },
             "data_types": [
                 "CRISPR_Achilles_Broad",
                 "Drug_CTD_Broad",
                 "Drug_Repurposing_Broad",
             ],
-            "all_depmap_ids": [[0, "ACH-000019"]],
+            "all_depmap_ids": [
+                [0, "ACH-000019"],
+                [1, "ACH-000020"],
+                [2, "ACH-000021"],
+                [3, "ACH-000022"],
+            ],
         }