test: support API for sequences

tests/conftest.py

@@ -1,8 +1,14 @@
 import os
 
+from Bio import SeqIO
+from Bio.SeqRecord import SeqRecord
+from Bio.Seq import Seq
+
 import numpy as np
 import pyBigWig
 import pytest
+import pyfaidx
+import random
 
 
 @pytest.fixture(scope="session")
@@ -11,6 +17,46 @@ def momics_path(tmp_path_factory):
     return p
 
 
+@pytest.fixture(scope="session")
+def fa1(tmp_path_factory):
+    p = os.path.join(tmp_path_factory.getbasetemp(), "fa1")
+    nucleotides = ["A", "T", "C", "G"]
+    chrom_seqs = {
+        "I": "".join(random.choices(nucleotides, k=10000)),
+        "II": "".join(random.choices(nucleotides, k=20000)),
+        "III": "".join(random.choices(nucleotides, k=30000)),
+    }
+    records = []
+    for chrom, sequence in chrom_seqs.items():
+        record = SeqRecord(Seq(sequence), id=chrom, description="")
+        records.append(record)
+
+    with open(p, "w") as output_handle:
+        SeqIO.write(records, output_handle, "fasta")
+
+    return p
+
+
+@pytest.fixture(scope="session")
+def fa2(tmp_path_factory):
+    p = os.path.join(tmp_path_factory.getbasetemp(), "fa2")
+    nucleotides = ["A", "T", "C", "G"]
+    chrom_seqs = {
+        "I": "".join(random.choices(nucleotides, k=2000)),
+        "II": "".join(random.choices(nucleotides, k=1000)),
+        "III": "".join(random.choices(nucleotides, k=500)),
+    }
+    records = []
+    for chrom, sequence in chrom_seqs.items():
+        record = SeqRecord(Seq(sequence), id=chrom, description="")
+        records.append(record)
+
+    with open(p, "w") as output_handle:
+        SeqIO.write(records, output_handle, "fasta")
+
+    return p
+
+
 @pytest.fixture(scope="session")
 def bw1(tmp_path_factory):
     p = os.path.join(tmp_path_factory.getbasetemp(), "bw1")

tests/test_api.py

@@ -72,15 +72,38 @@ def test_Momics_add_tracks(momics_path: str, bw1: str, bw2: str):
     print(mom.tracks())
 
 
-def test_Momics_query(momics_path: str):
+def test_Momics_add_seq(momics_path: str, fa1: str, fa2: str):
     mom = momics.Momics(momics_path, create=False)
-    q = mom.query("I:991-1010")
+
+    with pytest.raises(Exception, match=r".*do not have identical chromomosome.*"):
+        mom.add_sequence(fa2)
+
+    mom.add_sequence(fa1)
+
+    with pytest.raises(ValueError, match=r"Sequence already added to the repository"):
+        mom.add_sequence(fa2)
+
+    print(mom.sequence())
+
+
+def test_Momics_query_tracks(momics_path: str):
+    mom = momics.Momics(momics_path, create=False)
+    q = mom.query_tracks("I:991-1010")
     assert q.shape == (40, 3)
 
-    q = mom.query("I")
+    q = mom.query_tracks("I")
     assert q.shape == (20000, 3)
 
 
+def test_Momics_query_seqs(momics_path: str):
+    mom = momics.Momics(momics_path, create=False)
+    q = mom.query_sequence("I:991-1010")
+    assert q.shape == (20,)
+
+    q = mom.query_sequence("I")
+    assert q.shape == (10000,)
+
+
 def test_Momics_remove_tracks(momics_path: str, bw1: str, bw2: str):
     mom = momics.Momics(momics_path, create=False)
     mom.add_tracks({"bw3": bw1})
@@ -94,5 +117,5 @@ def test_Momics_remove_tracks(momics_path: str, bw1: str, bw2: str):
         }
     )
     assert mom.tracks().__eq__(out).all().all()
-    q = mom.query("I:991-1010")
+    q = mom.query_tracks("I:991-1010")
     assert np.unique(q["label"]).__eq__(["bw2", "bw3", "bw4"]).all()