merge master

Author: Hoeze

.circleci/config.yml

@@ -34,8 +34,8 @@ variables:
       name: Install conda dependencies
       command: |
         source activate kipoi-dev
-        apt update -y
-        apt install build-essential -y
+        apt-get --allow-releaseinfo-change update -y  
+        apt-get install build-essential -y
         conda install -y cython
         conda install -y -c bioconda cyvcf2 pybedtools pyfaidx biopython
   install_kipoi: &install_kipoi

kipoiseq/dataclasses.py

@@ -85,6 +85,12 @@ def start(self):
         """
         return self.pos - 1
 
+    @property
+    def end(self):
+        """1-based variant end position
+        """
+        return self.start + len(self.ref)
+
     @classmethod
     def from_cyvcf(cls, obj):
         if len(obj.ALT) > 1:

kipoiseq/dataloaders/sequence.py

@@ -1,5 +1,6 @@
 import pandas as pd
 import numpy as np
+import pyranges as pr
 from copy import deepcopy
 from kipoi.metadata import GenomicRanges
 from kipoi.data import Dataset, kipoi_dataloader
@@ -459,7 +460,6 @@ def __init__(self, gtf_file, fasta_file,
                  transform=one_hot_dna,
                  interval_attrs=["gene_id", "Strand"],
                  use_strand=True):
-        import pyranges as pr
 
         # Read and filter gtf
         gtf = pr.read_gtf(gtf_file).df

kipoiseq/extractors/vcf_matching.py

@@ -31,7 +31,7 @@ def variants_to_pyranges(variants: List[Variant]) -> PyRanges:
         (
             v.chrom,
             v.start,
-            v.start + max(len(v.ref), len(v.alt)),
+            v.end,
             v
         )
         for v in variants
@@ -271,7 +271,7 @@ def iter_rows(self):
                 for _, row in df.iterrows():
                     yield row
 
-    def __iter__(self):
+    def __iter__(self) -> (Interval, Variant):
         """
         Iterate interval and variant object.
         """

tests/extractors/test_vcf_matching.py

@@ -118,20 +118,18 @@ def test_SingleVariantMatcher__iter__():
     assert (inters[0], variants[0]) in pairs
     assert (inters[0], variants[1]) in pairs
     assert (inters[1], variants[2]) in pairs
-    assert (inters[2], variants[1]) in pairs
     assert (inters[2], variants[2]) in pairs
 
-    assert len(pairs) == 5
+    assert len(pairs) == 4
 
     matcher = SingleVariantMatcher(vcf_file, pranges=pr)
     pairs = list(matcher)
 
     assert (inters[0], variants[0]) in pairs
     assert (inters[0], variants[1]) in pairs
     assert (inters[1], variants[2]) in pairs
-    assert (inters[2], variants[1]) in pairs
     assert (inters[2], variants[2]) in pairs
-    assert len(pairs) == 5
+    assert len(pairs) == 4
 
     matcher = SingleVariantMatcher(variants=variants, pranges=pr)
     pairs = list(matcher)