issue #1059 - all variants node

analysis/forms/forms_nodes.py

@@ -165,7 +165,8 @@ def __init__(self, analysis: Analysis, *args, **kwargs):
 class AllVariantsNodeForm(BaseNodeForm):
     class Meta:
         model = AllVariantsNode
-        fields = ('max_variant', "gene_symbol", "reference",
+        fields = ('max_variant', "gene_symbol",
+                  "reference", "snps", "indels", "structural_variants",
                   "min_het_or_hom_count", "max_het_or_hom_count",
                   "min_unk_count", "max_unk_count", "min_ref_count", "max_ref_count",
                   "min_hom_count", "max_hom_count", "min_het_count", "max_het_count")

analysis/migrations/0085_allvariantsnode_complex_subsitution_and_more.py

@@ -0,0 +1,33 @@
+# Generated by Django 4.2.10 on 2024-05-10 07:54
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('analysis', '0084_remove_karyomappinggene_gene_and_more'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='allvariantsnode',
+            name='complex_subsitution',
+            field=models.BooleanField(blank=True, default=True),
+        ),
+        migrations.AddField(
+            model_name='allvariantsnode',
+            name='indels',
+            field=models.BooleanField(blank=True, default=True),
+        ),
+        migrations.AddField(
+            model_name='allvariantsnode',
+            name='snps',
+            field=models.BooleanField(blank=True, default=True),
+        ),
+        migrations.AddField(
+            model_name='allvariantsnode',
+            name='structural_variants',
+            field=models.BooleanField(blank=True, default=True),
+        ),
+    ]

analysis/models/nodes/sources/all_variants_node.py

@@ -1,4 +1,5 @@
-from functools import cached_property
+import operator
+from functools import cached_property, reduce
 from typing import Optional
 
 from django.db import models
@@ -15,6 +16,10 @@ class AllVariantsNode(AnalysisNode, AbstractZygosityCountNode):
     max_variant = models.ForeignKey(Variant, null=True, on_delete=SET_NULL)
     gene_symbol = models.ForeignKey(GeneSymbol, null=True, blank=True, on_delete=CASCADE)
     reference = models.BooleanField(default=False, blank=True)
+    snps = models.BooleanField(default=True, blank=True)
+    indels = models.BooleanField(default=True, blank=True)
+    complex_subsitution = models.BooleanField(default=True, blank=True)
+    structural_variants = models.BooleanField(default=True, blank=True)
     min_inputs = 0
     max_inputs = 0
 
@@ -43,9 +48,26 @@ def _get_node_arg_q_dict(self) -> dict[Optional[str], dict[str, Q]]:
             q_gene = Q(variantgeneoverlap__gene__in=genes)
             q_dict[str(q_gene)] = q_gene
 
-        if not self.reference:
-            q_no_ref = Variant.get_no_reference_q()
-            q_dict[str(q_no_ref)] = q_no_ref
+        q_lookup = [
+            (self.reference, Variant.get_reference_q()),
+            (self.snps, Variant.get_snp_q()),
+            (self.indels, Variant.get_indel_q()),
+            (self.complex_subsitution, Variant.get_complex_subsitution_q()),
+            (self.structural_variants, Variant.get_symbolic_q()),
+        ]
+
+        filters = []
+        all_true = True
+        for test, q in q_lookup:
+            all_true &= test
+            if test:
+                filters.append(q)
+
+        if not all_true:
+            if filters:
+                q_dict["variant_types"] = reduce(operator.or_, filters)
+            else:
+                q_dict["empty"] = Q(pk__isnull=False)
 
         arg_q_dict = {None: q_dict}
         self.merge_arg_q_dicts(arg_q_dict, self.get_zygosity_count_arg_q_dict())

analysis/templates/analysis/node_editors/allvariantsnode_editor.html

@@ -30,11 +30,31 @@
 			<label for="{{ form.gene_symbol.id_for_label }}">Gene:</label>
 			{{ form.gene_symbol }}
 		</span>
-		<span class="fieldWrapper">
-			{{ form.reference.errors }}
-			<label for="{{ form.reference.id_for_label }}">Show reference variants:</label>
-			{{ form.reference }}
-		</span>
+        <div>
+            <span class="fieldWrapper">
+                {{ form.reference.errors }}
+                <label for="{{ form.reference.id_for_label }}">Reference variants:</label>
+                {{ form.reference }}
+            </span>
+            |
+            <span class="fieldWrapper">
+                {{ form.snps.errors }}
+                <label for="{{ form.snps.id_for_label }}">SNPs:</label>
+                {{ form.snps }}
+            </span>
+            |
+            <span class="fieldWrapper">
+                {{ form.indels.errors }}
+                <label for="{{ form.indels.id_for_label }}">indels:</label>
+                {{ form.indels }}
+            </span>
+            |
+            <span class="fieldWrapper">
+                {{ form.structural_variants.errors }}
+                <label for="{{ form.structural_variants.id_for_label }}">Structural Variants:</label>
+                {{ form.structural_variants }}
+            </span>
+        </div>
 
 		<table id="zygosity-table">
 			<tr>

snpdb/models/models_variant.py

@@ -567,6 +567,9 @@ class Variant(PreviewModelMixin, models.Model):
         There is only 1 Variant for a given locus/alt per database (handled via insertion queues) """
 
     REFERENCE_ALT = "="
+    _BASES = "GATC"
+    _REGEX_2_PLUS_BASES = f"^[GATC]{2,}$"
+
     locus = models.ForeignKey(Locus, on_delete=CASCADE)
     alt = models.ForeignKey(Sequence, on_delete=CASCADE)
     # end is a calculated field, but we store as an optimisation for overlap queries (start = locus.position)
@@ -589,14 +592,38 @@ def get_contigs_q(genome_build: GenomeBuild) -> Q:
         """ Restrict to contigs in a genome build """
         return Q(locus__contig__genomebuildcontig__genome_build=genome_build)
 
+    @staticmethod
+    def get_reference_q() -> Q:
+        return Q(alt__seq=Variant.REFERENCE_ALT)
+
     @staticmethod
     def get_no_reference_q() -> Q:
-        return ~Q(alt__seq=Variant.REFERENCE_ALT)
+        return ~Variant.get_reference_q()
 
     @staticmethod
     def get_snp_q() -> Q:
-        bases = "GATC"
-        return Q(locus__ref__seq__in=bases) & Q(alt__seq__in=bases)
+        return Q(locus__ref__seq__in=Variant._BASES) & Q(alt__seq__in=Variant._BASES)
+
+    @staticmethod
+    def get_insertion_q() -> Q:
+        return Q(locus__ref__seq__in=Variant._BASES) & Q(alt__seq__regex=Variant._REGEX_2_PLUS_BASES)
+
+    @staticmethod
+    def get_deletion_q() -> Q:
+        return Q(locus__ref__seq__regex=Variant._REGEX_2_PLUS_BASES) & Q(alt__seq__in=Variant._BASES)
+
+    @staticmethod
+    def get_indel_q() -> Q:
+        return Variant.get_insertion_q() | Variant.get_deletion_q()
+
+    @staticmethod
+    def get_complex_subsitution_q() -> Q:
+        regex_2_plus_bases = f"^[GATC]{2,}$"
+        return Q(locus__ref__seq__regex=regex_2_plus_bases) & Q(alt__seq__regex=regex_2_plus_bases)
+
+    @staticmethod
+    def get_symbolic_q() -> Q:
+        return Q(locus__ref__seq__startswith="<") & Q(alt__seq__startswith="<")
 
     @staticmethod
     def annotate_variant_string(qs, name="variant_string", path_to_variant=""):