diff --git a/lib/kb_bedtools/kb_bedtoolsImpl.py b/lib/kb_bedtools/kb_bedtoolsImpl.py
index eb5b813..570788e 100644
--- a/lib/kb_bedtools/kb_bedtoolsImpl.py
+++ b/lib/kb_bedtools/kb_bedtoolsImpl.py
@@ -9,8 +9,7 @@
from installed_clients.ReadsUtilsClient import ReadsUtils
from base import Core
-from kb_bedtools.utils import Intersection
-from kb_bedtools.utils import BamConversion
+from kb_bedtools.utils import BamConversion, Intersection
#END_HEADER
diff --git a/lib/kb_bedtools/utils.py b/lib/kb_bedtools/utils.py
index 94d9dc0..16738fb 100644
--- a/lib/kb_bedtools/utils.py
+++ b/lib/kb_bedtools/utils.py
@@ -3,6 +3,8 @@
import logging
import os
import subprocess
+import uuid
+import shutil
from collections import Counter
from shutil import copyfile
@@ -45,7 +47,12 @@ def do_analysis(self, params: dict):
wsname = params['workspace_name']
sequencing_tech = 'Illumina'
interleaved = params['interleaved']
- fastq_path = self.bam_to_fastq(staging_path, shared_folder=self.shared_folder)
+ if params.get('paired_end'):
+ fastq_path = self.bam_to_paired_fastq(staging_path, shared_folder=self.shared_folder)
+ else:
+ fastq_path = self.bam_to_fastq(staging_path, shared_folder=self.shared_folder)
+
+
reads_result = self.upload_reads(output_name, fastq_path, wsname, sequencing_tech, interleaved)
@@ -69,27 +76,63 @@ def do_analysis(self, params: dict):
@classmethod
def bam_to_fastq(cls, bam_file, shared_folder=""): # add a dict parameter so those parameter could be use
- with open(bam_file, 'rb') as file:
- bam_data = file.read().decode('utf-8', 'ignore')
- # best to use logging here so that messages are more visible
+ if not os.path.isfile(bam_file):
+ raise FileNotFoundError(f"{bam_file} not found")
+
+ unique_id = str(uuid.uuid4())[:8]
+ temp_fastq = f"filename_end1_{unique_id}.fq"
+ output_path = os.path.join(shared_folder, f"output_{unique_id}.fq")
+
logging.warning(f'{">"*20}{os.getcwd()}')
+
with subprocess.Popen([
- 'bedtools', 'bamtofastq', '-i', bam_file, '-fq', 'filename_end1.fq'
+ 'bedtools', 'bamtofastq', '-i', bam_file, '-fq', temp_fastq
]) as proc:
proc.wait()
- if not os.path.exists("filename_end1.fq"):
- raise FileNotFoundError("bedtools did not create FASTQ file")
- if os.path.getsize("filename_end1.fq") < 100:
+ if not os.path.exists(temp_fastq):
+ raise FileNotFoundError("bedtools did not create FASTQ file")
+
+ if os.path.getsize(temp_fastq) < 100:
raise ValueError("Generated FASTQ file is unexpectedly small — check input BAM or bedtools error")
- output_path = os.path.join(shared_folder, 'output.fq')
- copyfile('filename_end1.fq', output_path)
-
+ shutil.copyfile(temp_fastq, output_path)
return output_path
-
- def upload_reads(self, name, reads_path, workspace_name, sequencing_tech, interleaved):
+ @classmethod
+ def bam_to_fastq_paired(cls, bam_file, shared_folder=""):
+ if not os.path.isfile(bam_file):
+ raise FileNotFoundError(f"{bam_file} not found")
+
+ unique_id = str(uuid.uuid4())[:8]
+ read1_file = f"filename_end1_{unique_id}.fq"
+ read2_file = f"filename_end2_{unique_id}.fq"
+
+
+ logging.warning(f"{'>'*20}{os.getcwd()}")
+
+ with subprocess.Popen([
+ 'bedtools', 'bamtofastq',
+ '-i', bam_file,
+ '-fq', read1_file,
+ '-fq2', read2_file
+ ]) as proc:
+ proc.wait()
+
+ if not os.path.exists(read1_file) or not os.path.exists(read2_file):
+ raise FileNotFoundError("Paired-end FASTQ files were not created")
+
+
+ output1 = os.path.join(shared_folder, f"read1_{unique_id}.fq")
+ output2 = os.path.join(shared_folder, f"read2_{unique_id}.fq")
+
+ shutil.copyfile(read1_file, output1)
+ shutil.copyfile(read2_file, output2)
+
+ return {"read1": output1, "read2": output2}
+
+ def upload_reads(self, name, reads_path, workspace_name,
+ sequencing_tech=None, interleaved=None):
"""
Upload reads back to the KBase Workspace. This method only uses the
minimal parameters necessary to provide a demonstration. There are many
diff --git a/test/kb_bedtools_server_test.py b/test/kb_bedtools_server_test.py
index 30fc07c..3434376 100644
--- a/test/kb_bedtools_server_test.py
+++ b/test/kb_bedtools_server_test.py
@@ -95,6 +95,7 @@ def test_intersect(self):
"output_name": "intersectOutput",
})
+ def test_bamtofastq(self):
params = {
"workspace_name": self.wsName,
"reads_ref": "70257/2/1",
diff --git a/test/minimal.bam b/test/minimal.bam
index 90a0dcd..8d7e3cf 100644
Binary files a/test/minimal.bam and b/test/minimal.bam differ
diff --git a/test/unit_tests/test_kb_bedtools_utils.py b/test/unit_tests/test_kb_bedtools_utils.py
index 78327d0..f367054 100644
--- a/test/unit_tests/test_kb_bedtools_utils.py
+++ b/test/unit_tests/test_kb_bedtools_utils.py
@@ -1,7 +1,8 @@
import logging
import subprocess
-
+import os
import pytest
+from kb_bedtools.utils import BamConversion, Intersection
@pytest.fixture
@@ -15,3 +16,71 @@ def process():
def test_process(process):
logging.info("Running a test")
assert 1 == 1
+
+def test_bam_to_fastq_creates_output(tmp_path):
+ output_dir = tmp_path / "output"
+ output_dir.mkdir()
+
+ bam_path = os.path.join(os.path.dirname(__file__), "..", "minimal.bam")
+
+ result_path = BamConversion.bam_to_fastq(bam_path, str(output_dir))
+
+ assert os.path.exists(result_path), "FASTQ output file was not created"
+ with open(result_path) as f:
+ contents = f.read()
+ assert contents.startswith("@"), "FASTQ output does not appear valid"
+
+def test_bam_to_fastq_invalid_input_raises():
+ with pytest.raises(Exception):
+ BamConversion.bam_to_fastq("/invalid/path/to/file.bam")
+
+def test_bam_to_fastq_empty_file_raises(tmp_path):
+ empty_bam = tmp_path / "empty.bam"
+ empty_bam.touch()
+
+ with pytest.raises(Exception):
+ BamConversion.bam_to_fastq(str(empty_bam), str(tmp_path))
+
+
+def test_bam_to_fastq_multiple_outputs(tmp_path):
+ bam_path = os.path.join(os.path.dirname(__file__), "..", "minimal.bam")
+
+ result1 = BamConversion.bam_to_fastq(bam_path, tmp_path)
+ result2 = BamConversion.bam_to_fastq(bam_path, tmp_path)
+
+ assert result1 != result2, "Multiple conversions should create unique output files"
+
+def test_bam_to_fastq_paired_end_output(tmp_path):
+ bam_path = os.path.join(os.path.dirname(__file__), "..", "minimal.bam")
+
+ result_paths = BamConversion.bam_to_fastq_paired(
+ bam_file=bam_path,
+ shared_folder=str(tmp_path)
+ )
+
+ try:
+
+ assert isinstance(result_paths, dict), "Output should be a dictionary"
+ assert "read1" in result_paths and "read2" in result_paths, "Missing paired-end keys"
+ assert os.path.exists(result_paths["read1"]), "FASTQ for read 1 not created"
+ assert os.path.exists(result_paths["read2"]), "FASTQ for read 2 not created"
+
+
+ with open(result_paths["read1"]) as f1, open(result_paths["read2"]) as f2:
+ content1 = f1.read(1)
+ content2 = f2.read(1)
+
+ if content1:
+ assert content1 == "@", "Read 1 output is not a valid FASTQ file"
+ else:
+ print("Warning: Read 1 FASTQ file is empty")
+
+ if content2:
+ assert content2 == "@", "Read 2 output is not a valid FASTQ file"
+ else:
+ print ("Warning: Read 2 FASTQ file is empty")
+
+ finally:
+ for path in result_paths.values():
+ if os.path.exists(path):
+ os.remove(path)
diff --git a/ui/narrative/methods/run_kb_bedtools/display.yaml b/ui/narrative/methods/run_kb_bedtools/display.yaml
index 1220db5..22a41e9 100644
--- a/ui/narrative/methods/run_kb_bedtools/display.yaml
+++ b/ui/narrative/methods/run_kb_bedtools/display.yaml
@@ -59,5 +59,6 @@ parameters :
long-hint : |
Select if reads file is interleaved
+
description : |
- Takes a BAM file and returns a FASTQ file.
\ No newline at end of file
+ Takes a BAM file and returns a FASTQ file.