Exemplo n.º 1
0
def test_intersect_15():
    """
    A : --------------         -------
            ------
    B :       -----          ----------------
    R1:       -----            -------      (overlap)
              ----
    R2: --------------         -------      (original)
            ------
    R3:                        -------      (comp_incl)
    """
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=50, name="test",
                        strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=20, end=40, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=70, end=80, name="test", strand="."))
    genset2 = GenCoorSet(name="Test_set")
    genset2.add(
        GenCoor(chrom="chr1", start=25, end=45, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=65, end=95, name="test", strand="."))
    res = genset1.intersect(genset2, mode="overlap")
    assert len(res) == 3
    res = genset1.intersect(genset2, mode="original")
    assert len(res) == 3
    res = genset1.intersect(genset2, mode="complete_included")
    assert len(res) == 1
Exemplo n.º 2
0
def test_intersect_8():
    """
    One overlapping region
    A : ------
    B :     --------
    R1:     --       (overlap)
    R2: ------       (original)
    R3:              (comp_incl)
    """
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=10, name="test",
                        strand="."))
    genset2 = GenCoorSet(name="Test_set")
    genset2.add(GenCoor(chrom="chr1", start=7, end=20, name="test",
                        strand="."))
    res = genset1.intersect(genset2, mode="overlap")
    assert len(res) == 1
    assert res[0].start == 7
    assert res[0].end == 10
    res = genset1.intersect(genset2, mode="original")
    assert len(res) == 1
    assert res[0].start == 1
    assert res[0].end == 10
    res = genset1.intersect(genset2, mode="complete_included")
    assert len(res) == 0
Exemplo n.º 3
0
def test_total_coverage1():
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=5, name="test", strand="."))
    genset1.add(GenCoor(chrom="chr2", start=2, end=4, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr3_random", start=1, end=80, name="test", strand="."))
    cov = genset1.total_coverage()
    assert cov == 85
Exemplo n.º 4
0
def test_distance2():
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=5, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=20, end=24, name="test", strand="."))
    g = GenCoor(chrom="chr1", start=7, end=19, name="test", strand=".")
    res = genset1.distance(g, sign=False)
    assert res == 1
Exemplo n.º 5
0
def test_intersect_10():
    """
    Two separately overlapping regions
    A : -------      --------
    B :     -----        --------
    R1:     ---          ----     (overlap)
    R2: -------      --------     (original)
    R3:                           (comp_incl)
    """
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=10, name="test",
                        strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=26, end=35, name="test", strand="."))
    genset2 = GenCoorSet(name="Test_set")
    genset2.add(GenCoor(chrom="chr1", start=7, end=15, name="test",
                        strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=30, end=40, name="test", strand="."))
    res = genset1.intersect(genset2, mode="overlap")
    assert len(res) == 2
    res = genset1.intersect(genset2, mode="original")
    assert len(res) == 2
    res = genset1.intersect(genset2, mode="complete_included")
    assert len(res) == 0
Exemplo n.º 6
0
def test_intersect_3():
    """
    A : none
    B :   -----
    R : none
    """
    genset1 = GenCoorSet(name="Test_set")
    genset2 = GenCoorSet(name="Test_set")
    genset2.add(
        GenCoor(chrom="chr1", start=10, end=20, name="test", strand="+"))
    res = genset1.intersect(genset2, mode="overlap")
    assert len(res) == 0
    res = genset1.intersect(genset2, mode="original")
    assert len(res) == 0
    res = genset1.intersect(genset2, mode="complete_included")
    assert len(res) == 0
Exemplo n.º 7
0
def test_distances1():
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=5, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=80, end=84, name="test", strand="."))
    genset2 = GenCoorSet(name="Test_set")
    genset2.add(
        GenCoor(chrom="chr1", start=10, end=15, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=60, end=64, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=160, end=164, name="test", strand="."))
    res = genset1.distances(genset2, sign=False)
    print(res)
    assert len(res) == 2
    assert res == [5, 16]
Exemplo n.º 8
0
def test_intersect_12():
    """
    Different chromosomes
    A : chr1  -------
    B : chr2  -------
    R : none
    """
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=10, name="test",
                        strand="."))
    genset2 = GenCoorSet(name="Test_set")
    genset2.add(GenCoor(chrom="chr2", start=1, end=10, name="test",
                        strand="."))
    res = genset1.intersect(genset2, mode="overlap")
    assert len(res) == 0
    res = genset1.intersect(genset2, mode="original")
    assert len(res) == 0
    res = genset1.intersect(genset2, mode="complete_included")
    assert len(res) == 0
Exemplo n.º 9
0
def test_extend():
    genset = GenCoorSet(name="Test_set")
    genset.add(GenCoor(chrom="chr1", start=10, end=20, name="test",
                       strand="+"))
    genset.add(GenCoor(chrom="chr1", start=15, end=50, name="test",
                       strand="-"))
    genset.add(
        GenCoor(chrom="chr2", start=100, end=200, name="test", strand="."))
    ngcs = genset.extend(mode="left", length=5)
    assert ngcs.list[0].start == 5
    assert ngcs.list[0].end == 20
    assert ngcs.list[1].start == 10
    assert ngcs.list[1].end == 50
    ngcs = genset.extend(mode="right", length=5)
    assert ngcs.list[0].start == 10
    assert ngcs.list[0].end == 25
    assert ngcs.list[1].start == 15
    assert ngcs.list[1].end == 55
    ngcs = genset.extend(mode="5end", length=5)
    assert ngcs.list[0].start == 5
    assert ngcs.list[0].end == 20
    assert ngcs.list[1].start == 15
    assert ngcs.list[1].end == 55
    ngcs = genset.extend(mode="3end", length=5)
    assert ngcs.list[0].start == 10
    assert ngcs.list[0].end == 25
    assert ngcs.list[1].start == 10
    assert ngcs.list[1].end == 50
    ngcs = genset.extend(mode="both", length=5)
    assert ngcs.list[0].start == 5
    assert ngcs.list[0].end == 25
    assert ngcs.list[1].start == 10
    assert ngcs.list[1].end == 55
Exemplo n.º 10
0
def test_GenCoorSet_len():
    genset = GenCoorSet(name="Test_set")
    genset.add(GenCoor(chrom="chr1", start=10, end=20, name="test",
                       strand="."))
    genset.add(GenCoor(chrom="chr1", start=15, end=50, name="test",
                       strand="."))
    genset.add(
        GenCoor(chrom="chr2", start=100, end=200, name="test", strand="."))
    assert len(genset) == 3
Exemplo n.º 11
0
def test_rm_duplicates2():
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=5, name="test", strand="."))
    genset1.add(GenCoor(chrom="chr2", start=2, end=4, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr3_random", start=1, end=80, name="test", strand="."))
    genset1.add(GenCoor(chrom="chr2", start=2, end=4, name="test2",
                        strand="-"))
    res = genset1.rm_duplicates(inplace=False)
    assert len(res) == 4
Exemplo n.º 12
0
def test_standard_chromosome1():
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=50, name="test",
                        strand="."))
    genset1.add(
        GenCoor(chrom="chr2", start=20, end=40, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr3_random", start=70, end=80, name="test",
                strand="."))
    genset1.standard_chromosome()
    assert len(genset1) == 2
Exemplo n.º 13
0
def test_intersect_6():
    """
    No length attach
    A : .      .
    B :    .   .
    R : none
    """
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=2, end=2, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=20, end=20, name="test", strand="."))
    genset2 = GenCoorSet(name="Test_set")
    genset2.add(GenCoor(chrom="chr1", start=5, end=5, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=20, end=20, name="test", strand="."))
    res = genset1.intersect(genset2, mode="overlap")
    assert len(res) == 1
    res = genset1.intersect(genset2, mode="original")
    assert len(res) == 1
    res = genset1.intersect(genset2, mode="complete_included")
    assert len(res) == 1
Exemplo n.º 14
0
def test_intersect_13():
    """
    Completely included overlapping
    A : ---------------------------
    B : ----    ------       -----------
    R1: ----    ------       ------      (overlap)
    R2: ---------------------------      (original)
    R3:                                  (comp_incl)
    """
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=50, name="test",
                        strand="."))
    genset2 = GenCoorSet(name="Test_set")
    genset2.add(GenCoor(chrom="chr1", start=1, end=5, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=10, end=19, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=45, end=60, name="test", strand="."))
    res = genset1.intersect(genset2, mode="overlap")
    assert len(res) == 3
    res = genset1.intersect(genset2, mode="original")
    assert len(res) == 1
    res = genset1.intersect(genset2, mode="complete_included")
    assert len(res) == 0
Exemplo n.º 15
0
def test_merge():
    genset = GenCoorSet(name="Test_set")
    genset.add(GenCoor(chrom="chr1", start=10, end=20, name="test",
                       strand="+"))
    genset.add(GenCoor(chrom="chr1", start=15, end=50, name="test",
                       strand="-"))
    genset.add(
        GenCoor(chrom="chr2", start=100, end=200, name="test", strand="."))
    res = genset.merge(w_return=True)
    assert len(res) == 2
    assert res[0].start == 10
    assert res[0].end == 50
    assert res[0].strand == "."
    res = genset.merge(w_return=True, strand_specific=True)
    assert len(res) == 3
    assert res[0].start == 10
    assert res[0].end == 20
    assert res[0].strand == "+"
Exemplo n.º 16
0
def test_intersect_5():
    """
    End-to-end attach
    A : ------      ------
    B :       ------
    R : none
    """
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=5, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=11, end=20, name="test", strand="."))
    genset2 = GenCoorSet(name="Test_set")
    genset2.add(GenCoor(chrom="chr1", start=5, end=11, name="test",
                        strand="."))
    res = genset1.intersect(genset2, mode="overlap")
    assert len(res) == 0
    res = genset1.intersect(genset2, mode="original")
    assert len(res) == 0
    res = genset1.intersect(genset2, mode="complete_included")
    assert len(res) == 0
Exemplo n.º 17
0
def test_intersect_11():
    """
    Many various overlapping (mixed)
    A :   ------------------            --------   ---------
    B : ----   -------    ------            ----------
    R1:   --   -------    --                ----   ---       (overlap)
    R2:   ------------------            --------   --------- (original)
    R3:                                                      (comp_incl)
    """
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=3, end=30, name="test",
                        strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=50, end=60, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=70, end=85, name="test", strand="."))
    genset2 = GenCoorSet(name="Test_set")
    genset2.add(GenCoor(chrom="chr1", start=1, end=5, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=10, end=19, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=27, end=35, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=55, end=75, name="test", strand="."))
    res = genset1.intersect(genset2, mode="overlap")
    assert len(res) == 5
    res = genset1.intersect(genset2, mode="original")
    assert len(res) == 3
    res = genset1.intersect(genset2, mode="complete_included")
    assert len(res) == 0
Exemplo n.º 18
0
def test_total_coverage2():
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=5, name="test", strand="."))
    cov = genset1.total_coverage()
    assert cov == 4
Exemplo n.º 19
0
def test_intersect_4():
    """
    No overlapping
    A : ------      ---------               -------
    B :        ----          ------  ------
    R : none
    """
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=5, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=11, end=20, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=33, end=38, name="test", strand="."))
    genset2 = GenCoorSet(name="Test_set")
    genset2.add(GenCoor(chrom="chr1", start=7, end=9, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=20, end=25, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=26, end=31, name="test", strand="."))
    res = genset1.intersect(genset2, mode="overlap")
    assert len(res) == 0
    res = genset1.intersect(genset2, mode="original")
    assert len(res) == 0
    res = genset1.intersect(genset2, mode="complete_included")
    assert len(res) == 0
Exemplo n.º 20
0
def test_intersect_7():
    """
    Perfect overlapping
    A : ------
    B : ------
    R : ------
    """
    genset1 = GenCoorSet(name="Test_set")
    genset1.add(GenCoor(chrom="chr1", start=1, end=10, name="test",
                        strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=500, end=550, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=600, end=650, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=700, end=750, name="test", strand="."))
    genset1.add(
        GenCoor(chrom="chr1", start=725, end=800, name="test", strand="."))
    genset2 = GenCoorSet(name="Test_set")
    genset2.add(GenCoor(chrom="chr1", start=1, end=10, name="test",
                        strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=500, end=550, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=600, end=650, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=700, end=750, name="test", strand="."))
    genset2.add(
        GenCoor(chrom="chr1", start=725, end=800, name="test", strand="."))
    res = genset1.intersect(genset2, mode="overlap")
    assert len(res) == 6
    res = genset1.intersect(genset2, mode="original")
    assert len(res) == 5
    res = genset1.intersect(genset2, mode="complete_included")
    assert len(res) == 5