feat: infer categories of categorical columns (#229)

feat: allow lineup to infer categories
Caleydo · Oct 13, 2023 · 71953f8 · 71953f8
1 parent 3690eea
commit 71953f8
Show file tree

Hide file tree

Showing 2 changed files with 29 additions and 29 deletions.
diff --git a/src/common/config.ts b/src/common/config.ts
@@ -185,9 +185,7 @@ export const tissue: IDataSourceConfig = {
   },
 };
 
-function toChromosomes(categories: (string | Partial<ICategory>)[]) {
-  const mappedCategories: Partial<ICategory>[] = categories.map((category) => (typeof category === 'string' ? { name: category, label: category } : category));
-
+function toChromosomes(categories: Partial<ICategory>[]) {
   const order = new Map<string, number>();
   for (let i = 1; i <= 22; ++i) {
     order.set(String(i), i);
@@ -196,7 +194,7 @@ function toChromosomes(categories: (string | Partial<ICategory>)[]) {
   order.set('y', 24);
   order.set('mt', 25);
 
-  mappedCategories.sort((a, b) => {
+  categories.sort((a, b) => {
     const an = a.label.toLowerCase();
     const bn = b.label.toLowerCase();
     const ai = order.get(an);
@@ -213,7 +211,7 @@ function toChromosomes(categories: (string | Partial<ICategory>)[]) {
     return ai - bi;
   });
 
-  return mappedCategories.map((d, i) => ({ name: d.name, label: d.label, value: i }));
+  return categories;
 }
 
 export const gene: IDataSourceConfig = {
@@ -231,7 +229,10 @@ export const gene: IDataSourceConfig = {
       ColumnDescUtils.stringCol('symbol', { label: 'Symbol', width: 120 }),
       ColumnDescUtils.stringCol('id', { label: 'Ensembl' }),
       ColumnDescUtils.stringCol('name', { label: 'Name' }),
-      ColumnDescUtils.categoricalCol('chromosome', toChromosomes(find('chromosome').categories), { label: 'Chromosome' }),
+      ColumnDescUtils.categoricalCol('chromosome', find('chromosome').categories, {
+        label: 'Chromosome',
+        extras: { categoryOrder: toChromosomes },
+      }),
       ColumnDescUtils.categoricalCol('biotype', find('biotype').categories, { label: 'Biotype' }),
       ColumnDescUtils.categoricalCol(
         'strand',

diff --git a/tdp_publicdb/entity.py b/tdp_publicdb/entity.py
@@ -16,7 +16,7 @@
 
 
 def _gene_columns(query):
-    return query.column("biotype", type="categorical").column("chromosome", type="categorical")
+    return query.column("biotype", type="categorical", categories=None).column("chromosome", type="categorical", categories=None)
 
 
 gene = Entity(
@@ -37,31 +37,30 @@ def _gene_columns(query):
 
 def _tissue_columns(query):
     return (
-        query.column("species", type="categorical")
-        .column("tumortype", type="categorical")
-        .column("organ", type="categorical")
-        .column("gender", type="categorical")
+        query.column("species", type="categorical", categories=None)
+        .column("tumortype", type="categorical", categories=None)
+        .column("organ", type="categorical", categories=None)
+        .column("gender", type="categorical", categories=None)
         .column("tumortype_adjacent", type="string")
-        .column("vendorname", type="categorical")
-        .column("race", type="categorical")
-        .column("ethnicity", type="categorical")
+        .column("vendorname", type="categorical", categories=None)
+        .column("race", type="categorical", categories=None)
+        .column("ethnicity", type="categorical", categories=None)
         .column("age", type="number")
         .column("days_to_death", type="number")
         .column("days_to_last_followup", type="number")
         .column(
             "vital_status",
-            type="categorical",
             categories=[dict(name="true", label="Alive", color="white"), dict(name="false", label="Deceased", color="black")],
         )
         .column("height", type="number")
         .column("weight", type="number")
         .column("bmi", type="number")
         .column("tumorpurity", type="number")
         .column("microsatellite_stability_score", type="number")
-        .column("microsatellite_stability_class", type="categorical")
+        .column("microsatellite_stability_class", type="categorical", categories=None)
         .column("mutational_fraction", type="number")
-        .column("hla_a_allele1", type="categorical")
-        .column("hla_a_allele2", type="categorical")
+        .column("hla_a_allele1", type="categorical", categories=None)
+        .column("hla_a_allele2", type="categorical", categories=None)
     )
 
 
@@ -106,20 +105,20 @@ def _tissue_columns(query):
 
 def _cellline_columns(query):
     return (
-        query.column("tumortype", type="categorical")
-        .column("organ", type="categorical")
-        .column("gender", type="categorical")
-        .column("metastatic_site", type="categorical")
-        .column("histology_type", type="categorical")
-        .column("morphology", type="categorical")
-        .column("growth_type", type="categorical")
-        .column("age_at_surgery", type="categorical")
+        query.column("tumortype", type="categorical", categories=None)
+        .column("organ", type="categorical", categories=None)
+        .column("gender", type="categorical", categories=None)
+        .column("metastatic_site", type="categorical", categories=None)
+        .column("histology_type", type="categorical", categories=None)
+        .column("morphology", type="categorical", categories=None)
+        .column("growth_type", type="categorical", categories=None)
+        .column("age_at_surgery", type="categorical", categories=None)
         .column("cosmicid", type="number")
         .column("mutational_fraction", type="number")
         .column("microsatellite_stability_score", type="number")
-        .column("microsatellite_stability_class", type="categorical")
-        .column("hla_a_allele1", type="categorical")
-        .column("hla_a_allele2", type="categorical")
+        .column("microsatellite_stability_class", type="categorical", categories=None)
+        .column("hla_a_allele1", type="categorical", categories=None)
+        .column("hla_a_allele2", type="categorical", categories=None)
     )