Added some tests for the rust functions

pydeeptools/deeptools/test/test_data/log2ratio.bed

@@ -0,0 +1,3 @@
+3R	0	500	2.1154773
+3R	500	1000	1.4720684
+3R	1000	1500	0

pydeeptools/deeptools/test/test_r_alignmentsieve.py

@@ -0,0 +1,178 @@
+import pytest
+from deeptools.alignmentSieve2 import r_alignmentsieve
+import os.path
+
+
+ROOT = os.path.dirname(os.path.abspath(__file__)) + "/test_data/"
+BAMFILE_A = ROOT + "test_paired.bam"
+BAMFILE_B = ROOT + "test_paired2.bam"
+
+def test_r_alignmentsieve_basic():
+    bamifile = BAMFILE_A
+    ofile = BAMFILE_B
+    nproc = 4
+    filter_metrics = ROOT + "filter_metrics_basic.txt"
+    filtered_out_readsfile = ROOT + "filtered_out_reads.bam"
+    verbose = False
+    shift = []
+    _bed = False
+    filter_rna_strand = "None"
+    min_mapping_quality = 30
+    sam_flag_incl = 0
+    sam_flag_excl = 0
+    _blacklist = ""
+    min_fragment_length = 0
+    max_fragment_length = 1000
+    extend_reads = 0
+    center_reads = False
+
+    result = r_alignmentsieve(
+        bamifile, ofile, nproc, filter_metrics, filtered_out_readsfile, verbose, shift, _bed,
+        filter_rna_strand, min_mapping_quality, sam_flag_incl, sam_flag_excl, _blacklist,
+        min_fragment_length, max_fragment_length, extend_reads, center_reads
+    )
+
+    expected_content = [
+        "#bamFilterReads --filterMetrics",
+        "#File\tReads\tRemaining Total\tInitial Reads",
+        "test_paired.bam\t49\t49"
+    ]
+
+    with open(filter_metrics, 'r') as f:
+        lines = f.readlines()
+
+    # Remove the file path prefix from the third line
+    lines[2] = os.path.basename(lines[2])
+
+    # Strip newline characters from the lines
+    lines = [line.strip() for line in lines]
+
+    assert lines == expected_content
+
+
+def test_r_alignmentsieve_with_shift():
+    bamifile = BAMFILE_A
+    ofile = BAMFILE_B
+    nproc = 4
+    filter_metrics = ROOT + "filter_metrics_shift.txt"
+    filtered_out_readsfile = ROOT + "filtered_out_reads.bam"
+    verbose = False
+    shift = []
+    _bed = False
+    filter_rna_strand = "None"
+    min_mapping_quality = 30
+    sam_flag_incl = 0
+    sam_flag_excl = 0
+    _blacklist = ""
+    min_fragment_length = 0
+    max_fragment_length = 1000
+    extend_reads = 0
+    center_reads = False
+
+    result = r_alignmentsieve(
+        bamifile, ofile, nproc, filter_metrics, filtered_out_readsfile, verbose, shift, _bed,
+        filter_rna_strand, min_mapping_quality, sam_flag_incl, sam_flag_excl, _blacklist,
+        min_fragment_length, max_fragment_length, extend_reads, center_reads
+    )
+
+    expected_content = [
+        "#bamFilterReads --filterMetrics",
+        "#File\tReads\tRemaining Total\tInitial Reads",
+        "test_paired.bam\t49\t49"
+    ]
+
+    with open(filter_metrics, 'r') as f:
+        lines = f.readlines()
+
+    # Remove the file path prefix from the third line
+    lines[2] = os.path.basename(lines[2])
+
+    # Strip newline characters from the lines
+    lines = [line.strip() for line in lines]
+
+    assert lines == expected_content
+
+
+def test_r_alignmentsieve_with_filtering():
+    bamifile = BAMFILE_A
+    ofile = BAMFILE_B
+    nproc = 4
+    filter_metrics = ROOT + "filter_metrics_withFiltering.txt"
+    filtered_out_readsfile = ROOT +"filtered_out_reads.bam"
+    verbose = False
+    shift = []
+    _bed = False
+    filter_rna_strand = "forward"
+    min_mapping_quality = 30
+    sam_flag_incl = 0
+    sam_flag_excl = 0
+    _blacklist = ""
+    min_fragment_length = 100
+    max_fragment_length = 500
+    extend_reads = 0
+    center_reads = False
+
+    result = r_alignmentsieve(
+        bamifile, ofile, nproc, filter_metrics, filtered_out_readsfile, verbose, shift, _bed,
+        filter_rna_strand, min_mapping_quality, sam_flag_incl, sam_flag_excl, _blacklist,
+        min_fragment_length, max_fragment_length, extend_reads, center_reads
+    )
+
+    expected_content = [
+        "#bamFilterReads --filterMetrics",
+        "#File\tReads\tRemaining Total\tInitial Reads",
+        "test_paired.bam\t27\t49"
+    ]
+
+    with open(filter_metrics, 'r') as f:
+        lines = f.readlines()
+
+    # Remove the file path prefix from the third line
+    lines[2] = os.path.basename(lines[2])
+
+    # Strip newline characters from the lines
+    lines = [line.strip() for line in lines]
+
+    assert lines == expected_content
+
+def test_r_alignmentsieve_with_filtering_extendReads():
+    bamifile = BAMFILE_A
+    ofile = BAMFILE_B
+    nproc = 4
+    filter_metrics = ROOT + "filter_metrics_extendReads.txt"
+    filtered_out_readsfile = ROOT +"filtered_out_reads.bam"
+    verbose = False
+    shift = []
+    _bed = False
+    filter_rna_strand = "forward"
+    min_mapping_quality = 30
+    sam_flag_incl = 0
+    sam_flag_excl = 0
+    _blacklist = ""
+    min_fragment_length = 100
+    max_fragment_length = 500
+    extend_reads = 100
+    center_reads = False
+
+    result = r_alignmentsieve(
+        bamifile, ofile, nproc, filter_metrics, filtered_out_readsfile, verbose, shift, _bed,
+        filter_rna_strand, min_mapping_quality, sam_flag_incl, sam_flag_excl, _blacklist,
+        min_fragment_length, max_fragment_length, extend_reads, center_reads
+    )
+
+    expected_content = [
+        "#bamFilterReads --filterMetrics",
+        "#File\tReads\tRemaining Total\tInitial Reads",
+        "test_paired.bam\t27\t49"
+    ]
+
+    with open(filter_metrics, 'r') as f:
+        lines = f.readlines()
+
+    # Remove the file path prefix from the third line
+    lines[2] = os.path.basename(lines[2])
+
+    # Strip newline characters from the lines
+    lines = [line.strip() for line in lines]
+
+    assert lines == expected_content

pydeeptools/deeptools/test/test_r_bamCompare.py

@@ -0,0 +1,88 @@
+import pytest
+from deeptools.bamCompare2 import r_bamcompare
+import os.path
+import filecmp
+from os import unlink
+
+ROOT = os.path.dirname(os.path.abspath(__file__)) + "/test_data/"
+BAMFILE_A = ROOT + "test1.bam"
+BAMFILE_B = ROOT + "test2.bam"
+
+
+def test_r_bamcompare():
+    bamifile1 = BAMFILE_A
+    bamifile2 = BAMFILE_B
+    ofile = ROOT + "r_bamcompare_output.bedgraph"
+    ofiletype  = "bedgraph"
+    norm  = "None"
+    effective_genome_size = 0
+    scalefactorsmethod = "None"
+    operation = "ratio"
+    pseudocount = 0
+    # filtering options
+    ignoreduplicates = False
+    minmappingquality = 0
+    samflaginclude = 0
+    samflagexclude = 0
+    minfraglen = 0
+    maxfraglen = 0
+    nproc = 1
+    _ignorechr = []
+    binsize = 500
+    regions = []
+    verbose = False
+
+    # Call the Rust function
+    r_bamcompare(
+        bamifile1, bamifile2, ofile, ofiletype, norm, effective_genome_size, scalefactorsmethod, operation, 
+        pseudocount, ignoreduplicates, minmappingquality, samflaginclude, samflagexclude, minfraglen, maxfraglen, nproc, 
+        _ignorechr, binsize, regions, verbose
+    )
+    # Add assertions to verify the expected behavior
+    expected = ['3R\t0\t500\t4.571429\n',
+            '3R\t500\t1000\t2.8333333\n',
+            '3R\t1000\t1500\t1\n',]
+    _foo = open(ofile, 'r')
+    resp = _foo.readlines()
+    _foo.close()
+    assert f"{resp}" == f"{expected}", f"{resp} != {expected}"
+    unlink(ofile)
+
+def test_r_bamcompare_RPKM():
+    bamifile1 = BAMFILE_A
+    bamifile2 = BAMFILE_B
+    ofile = ROOT + "r_bamcompare_output.bedgraph"
+    ofiletype  = "bedgraph"
+    norm  = "RPKM"
+    effective_genome_size = 0
+    scalefactorsmethod = "None"
+    operation = "ratio"
+    pseudocount = 0
+    # filtering options
+    ignoreduplicates = False
+    minmappingquality = 0
+    samflaginclude = 0
+    samflagexclude = 0
+    minfraglen = 0
+    maxfraglen = 0
+    nproc = 1
+    _ignorechr = []
+    binsize = 500
+    regions = []
+    verbose = False
+
+    # Call the Rust function
+    r_bamcompare(
+        bamifile1, bamifile2, ofile, ofiletype, norm, effective_genome_size, scalefactorsmethod, operation, 
+        pseudocount, ignoreduplicates, minmappingquality, samflaginclude, samflagexclude, minfraglen, maxfraglen, nproc, 
+        _ignorechr, binsize, regions, verbose
+    )
+    # Add assertions to verify the expected behavior
+    expected = ['3R\t0\t500\t4.571429\n',
+            '3R\t500\t1000\t2.8333333\n',
+            '3R\t1000\t1500\t1\n',]
+    _foo = open(ofile, 'r')
+    resp = _foo.readlines()
+    _foo.close()
+    assert f"{resp}" == f"{expected}", f"{resp} != {expected}"
+    unlink(ofile)