Bioframe tests

README.md

@@ -1 +1,19 @@
 # polars_bio
+
+
+## Features
+
+| Features   | Bioframe | polars-bio | PyRanges | Pybedtools |
+|------------|----------|------------|----------|------------|
+| overlap    |:white_check_mark:|:white_check_mark:|          |            |
+| closest    |:white_check_mark:|            |          |            |
+| cluster    |:white_check_mark:|            |          |            |
+| complement |:white_check_mark:|            |          |            |
+
+## Input/Output
+| I/O              | Bioframe    | polars-bio        | PyRanges | Pybedtools |
+|------------------|-------------|-------------------|----------|------------|
+| Pandas DataFrame |:white_check_mark:|:white_check_mark:    |          |            |
+| Polars DataFrame |             |:white_check_mark:       |          |            |
+| Polars LazyFrame |             |:white_check_mark:       |          |            |
+| Native readers   |             |:white_check_mark:       |          |            |

tests/_expected.py

@@ -42,6 +42,10 @@
 PD_DF1 = pd.read_csv(DF_PATH1)
 PD_DF2 = pd.read_csv(DF_PATH2)
 
+BIO_PD_DF1 = pd.read_parquet(f"{DATA_DIR}/exons/")
+BIO_PD_DF2 = pd.read_parquet(f"{DATA_DIR}/fBrain-DS14718/")
+
+
 # Polars
 PL_DF_OVERLAP = pl.DataFrame(PD_DF_OVERLAP)
 PL_DF1 = pl.DataFrame(PD_DF1)

tests/test_bioframe.py

@@ -0,0 +1,22 @@
+import bioframe as bf
+import pandas as pd
+import polars_bio as pb
+
+from _expected import BIO_PD_DF1, BIO_PD_DF2
+from polars_bio import OverlapFilter
+
+class TestOverlapBioframe:
+    result = pb.overlap(BIO_PD_DF1, BIO_PD_DF2, output_type="pandas.DataFrame", overlap_filter=OverlapFilter.Strict)
+    result_bio = bf.overlap(BIO_PD_DF1, BIO_PD_DF2,
+                            cols1=('contig','pos_start','pos_end'),
+                            cols2=('contig','pos_start','pos_end'),
+                            suffixes=('_1', '_2'),
+                            how="inner")
+    def test_overlap_count(self):
+        assert len(self.result) == 54246
+        assert len(self.result) ==  len(self.result_bio)
+
+    def test_overlap_schema_rows(self):
+        expected = self.result_bio.sort_values(by=list(self.result.columns)).reset_index(drop=True)
+        result = self.result.sort_values(by=list(self.result.columns)).reset_index(drop=True)
+        pd.testing.assert_frame_equal(result, expected)