| """Keep the corrected UTR SFT mapping separate from scalar regression inputs.""" |
| import hashlib |
| import importlib.util |
| from pathlib import Path |
| import unittest |
|
|
| ROOT = Path(__file__).resolve().parents[1] |
| spec = importlib.util.spec_from_file_location('sft_mapping', ROOT / 'train_sft.py') |
| sft = importlib.util.module_from_spec(spec) |
| spec.loader.exec_module(sft) |
|
|
|
|
| class SFTDataMappingTests(unittest.TestCase): |
| def test_utr_wet_uses_fixed_utr_generation_data(self): |
| preset = sft.PROJECTS['utr-wet'] |
| self.assertEqual(preset['directory'], 'utrgen') |
| self.assertEqual(preset['length'], 512) |
| self.assertEqual(preset['batch'], 2) |
| self.assertEqual(preset['steps'], -1) |
| expected = { |
| 'train.txt': '48f7e5986a4d5816bdc5f5dc8cab2f2acdc8ba6872879c9d3fff0799a49def4a', |
| 'valid.txt': '139b7462887a63eabacb7273948bf896699eb5a1668b9a9381d595646efe0bb1', |
| } |
| for name, digest in expected.items(): |
| path = sft.local_path(preset['directory'] + '/' + name) |
| self.assertEqual(hashlib.sha256(path.read_bytes()).hexdigest(), digest) |
| with path.open() as stream: |
| self.assertTrue(all(line.startswith('~$predict_utr') for line in stream if line.strip())) |
| self.assertEqual(path.read_bytes(), (ROOT / 'utrgen' / name).read_bytes()) |
|
|
| def test_ribozyme_wet_mapping_is_unchanged(self): |
| self.assertEqual(sft.PROJECTS['ribozyme-wet']['directory'], 'ribozymegen-figure7/filteredwet') |
|
|
