def test_assembly_of_many_subsequences():
original_prefix = "ACTGAACCTTGGAAACCCTTTGGG"
original_allele = "CCCTTT"
original_suffix = "GGAAGGAAGGAATTTTTTTT"
# generate 100 subsequences of all combinations of 0-9
# characters trimmed from beginning of prefix vs. end of suffix
subsequences = [
VariantSequence(
prefix=original_prefix[i:],
alt=original_allele,
suffix=original_suffix[:-j] if j > 0 else original_suffix,
reads={str(i) + "_" + str(j)}) for i in range(10)
for j in range(10)
]
eq_(100, len(subsequences))
# adding one decoy sequence which doesn't match
decoy = VariantSequence(prefix="G" + original_prefix[1:],
alt=original_allele,
suffix=original_suffix,
reads={"decoy"})
input_sequences = subsequences + [decoy]
results = iterative_overlap_assembly(input_sequences,
min_overlap_size=len(original_allele))
eq_(len(results), 2)
result = results[0]
eq_(result.prefix, original_prefix)
eq_(result.alt, original_allele)
eq_(result.suffix, original_suffix)
eq_(len(result.reads), len(subsequences))
result_decoy = results[1]
eq_(result_decoy.sequence, decoy.sequence)
def test_variant_sequence_overlaps():
# AAA|GG|TT
vs_3A = VariantSequence(
prefix="AAA",
alt="GG",
suffix="TT",
reads=[
AlleleRead(
prefix="AAA", allele="GG", suffix="TT", name="1")])
# AA|GG|TT
vs_2A = VariantSequence(
prefix="AA",
alt="GG",
suffix="TT",
reads=[
AlleleRead(
prefix="AA", allele="GG", suffix="TT", name="1")])
for min_overlap_size in [1, 2, 3, 4, 5, 6]:
assert vs_3A.left_overlaps(vs_2A, min_overlap_size=min_overlap_size), \
"Expected %s to overlap %s from left (min overlap size=%d)" % (
vs_3A, vs_2A, min_overlap_size)
assert not vs_2A.left_overlaps(vs_3A, min_overlap_size=min_overlap_size), \
"Expected %s to not overlap %s from left (min overlap size=%d)" % (
vs_2A, vs_3A, min_overlap_size)
assert not vs_3A.left_overlaps(vs_2A, min_overlap_size=7), \
"Unexpected overlap between %s and %s for min_overlap_size=7" % (
vs_3A, vs_2A)
def test_variant_sequence_add_reads():
vs = VariantSequence(prefix="A", alt="C", suffix="G", reads={"1"})
# adding reads '2' and '3', sometimes multiple times
vs_result = vs.add_reads("2").add_reads("1").add_reads("2").add_reads("3")
expected = VariantSequence(prefix="A",
alt="C",
suffix="G",
reads={"1", "2", "3"})
eq_(vs_result, expected)
def test_variant_sequence_combine():
vs1 = VariantSequence(prefix="A", alt="C", suffix="GG", reads={"1"})
vs2 = VariantSequence(prefix="AA", alt="C", suffix="GG", reads={"2"})
vs_result_1_to_2 = vs1.combine(vs2)
expected = VariantSequence(prefix="AA", alt="C", suffix="GG", reads={"1", "2"})
eq_(vs_result_1_to_2, expected)
# shouldn't matter which sequence is first as an argument to the combine
# function
vs_result_2_to_1 = vs2.combine(vs1)
eq_(vs_result_2_to_1, expected)
def test_variant_sequence_min_coverage():
# 1: AA|C|TT
# 2: AA|C|T
# 3: A|C|TT
reads = [
AlleleRead(prefix="AA", allele="C", suffix="TT", name="1"),
AlleleRead(prefix="AA", allele="C", suffix="T", name="2"),
AlleleRead(prefix="A", allele="C", suffix="TT", name="3")
]
vs = VariantSequence(prefix="AA", alt="C", suffix="TT", reads=reads)
eq_(vs.min_coverage(), 2)
def test_variant_sequence_trim_by_coverage():
reads = [
AlleleRead(prefix="AA", allele="C", suffix="T", name="1"),
AlleleRead(prefix="A", allele="C", suffix="T", name="2")
]
vs = VariantSequence(prefix="AA", alt="C", suffix="T", reads=reads)
# every nucleotide is spanned by one read
eq_(vs.trim_by_coverage(1), vs)
vs_expected_trim_by_2 = VariantSequence(prefix="A",
alt="C",
suffix="T",
reads=reads)
eq_(vs.trim_by_coverage(2), vs_expected_trim_by_2)
def test_variant_sequence_mean_coverage():
# 1: AA|C|TT
# 2: AA|C|T
# 3: A|C|TT
reads = [
AlleleRead(prefix="AA", allele="C", suffix="TT", name="1"),
AlleleRead(prefix="AA", allele="C", suffix="T", name="2"),
AlleleRead(prefix="A", allele="C", suffix="TT", name="3")
]
vs = VariantSequence(prefix="AA", alt="C", suffix="TT", reads=reads)
# count the number of times a nucleotide in the sequences above
# is contained in a read
expected_mean_coverage = (2 + 3 + 3 + 3 + 2) / 5
eq_(vs.mean_coverage(), expected_mean_coverage)
def test_assembly_time():
original_prefix = "ACTGAACCTTGGAAACCCTTTGGG"
original_allele = "CCCTTT"
original_suffix = "GGAAGGAAGGAATTTTTTTTGGCC"
# generate 400 subsequences of all combinations of 0-19
# characters trimmed from beginning of prefix vs. end of suffix
subsequences = [
VariantSequence(
prefix=original_prefix[i:],
alt=original_allele,
suffix=original_suffix[:-j] if j > 0 else original_suffix,
reads={str(i) + "_" + str(j)}) for i in range(20)
for j in range(20)
]
eq_(len(subsequences), 400)
t_start = time()
results = iterative_overlap_assembly(subsequences,
min_overlap_size=len(original_allele))
t_end = time()
eq_(len(results), 1)
result = results[0]
eq_(result.prefix, original_prefix)
eq_(result.suffix, original_suffix)
t_elapsed = t_end - t_start
assert t_elapsed < 0.1, \
"Expected assembly of 400 sequences to take less than 100ms: %0.4fms" % (
t_elapsed * 1000,)
def test_variant_sequence_min_coverage():
# 1: AA|C|TT
# 2: AA|C|T
# 3: A|C|TT
reads = [
AlleleRead(
prefix="AA", allele="C", suffix="TT", name="1"),
AlleleRead(
prefix="AA", allele="C", suffix="T", name="2"),
AlleleRead(
prefix="A", allele="C", suffix="TT", name="3")
]
vs = VariantSequence(
prefix="AA",
alt="C",
suffix="TT",
reads=reads)
eq_(vs.min_coverage(), 2)
def test_collapse_substrings():
# AAA|C|GGG
vs_longer = VariantSequence(prefix="AAA",
alt="C",
suffix="GGG",
reads={"1"})
# AAA|C|GG
vs_shorter = VariantSequence(prefix="AAA",
alt="C",
suffix="GG",
reads={"2"})
vs_unrelated = VariantSequence("TAA", alt="C", suffix="GG", reads={"3"})
results = collapse_substrings([vs_longer, vs_shorter, vs_unrelated])
eq_(len(results),
2), "Expected two sequences, got %d: %s" % (len(results), results)
vs_combined = vs_longer.add_reads({"2"})
assert vs_combined in results, "Expected %s to be in %s" % (vs_combined,
results)
assert vs_unrelated in results, "Expected %s to be in %s" % (vs_unrelated,
results)
def test_variant_sequence_trim_by_coverage():
reads = [
AlleleRead(
prefix="AA", allele="C", suffix="T", name="1"),
AlleleRead(
prefix="A", allele="C", suffix="T", name="2")
]
vs = VariantSequence(
prefix="AA",
alt="C",
suffix="T",
reads=reads)
# every nucleotide is spanned by one read
eq_(vs.trim_by_coverage(1), vs)
vs_expected_trim_by_2 = VariantSequence(
prefix="A",
alt="C",
suffix="T",
reads=reads)
eq_(vs.trim_by_coverage(2), vs_expected_trim_by_2)
def test_variant_sequence_mean_coverage():
# 1: AA|C|TT
# 2: AA|C|T
# 3: A|C|TT
reads = [
AlleleRead(
prefix="AA", allele="C", suffix="TT", name="1"),
AlleleRead(
prefix="AA", allele="C", suffix="T", name="2"),
AlleleRead(
prefix="A", allele="C", suffix="TT", name="3")
]
vs = VariantSequence(
prefix="AA",
alt="C",
suffix="TT",
reads=reads)
# count the number of times a nucleotide in the sequences above
# is contained in a read
expected_mean_coverage = (2 + 3 + 3 + 3 + 2) / 5
eq_(vs.mean_coverage(), expected_mean_coverage)
def test_assembly_of_simple_sequence_from_mock_reads():
# Read sequences:
# AAAAA|CC|TTTTT
# AAAAA|CC|TTTTT
# GAAAAA|CC|TTTTTG
# AAAA|CC|TTTT
reads = [
# two identical reads with sequence AAAAA|CC|TTTTT
AlleleRead(prefix="A" * 5, allele="CC", suffix="T" * 5, name="dup1"),
AlleleRead(prefix="A" * 5, allele="CC", suffix="T" * 5, name="dup2"),
# longer sequence GAAAAA|CC|TTTTTG
AlleleRead(prefix="G" + "A" * 5,
allele="CC",
suffix="T" * 5 + "G",
name="longer"),
# shorter sequence AAAA|CC|TTTT
AlleleRead(prefix="A" * 4, allele="CC", suffix="T" * 4,
name="shorter"),
]
expected_variant_sequence = VariantSequence(prefix="G" + "A" * 5,
alt="CC",
suffix="T" * 5 + "G",
reads=reads)
initial_variant_sequences = initial_variant_sequences_from_reads(reads)
# expecting one fewer sequence than reads since two of the reads are
# duplicates
eq_(len(initial_variant_sequences), len(reads) - 1)
# calling into either iterative_overlap_assembly or greedy_merge should
# give same results
for fn in [greedy_merge, iterative_overlap_assembly]:
assembled_variant_sequences = fn(initial_variant_sequences,
min_overlap_size=1)
# since no reads contradict each other then we should get back a single
# assembled sequence
eq_(
len(assembled_variant_sequences), 1,
"Unexpected number of variant sequences: %s" %
(assembled_variant_sequences, ))
assembled_variant_sequence = assembled_variant_sequences[0]
eq_(assembled_variant_sequence, expected_variant_sequence)
eq_(len(assembled_variant_sequence.reads), len(reads))
eq_(assembled_variant_sequence.min_coverage(), 1)
# 2 bases with 1/4 reads, 2 bases with 3/4 reads, remaining 10 bases with
# all 4/4 reads
expected_mean_coverage = (2 * 1 + 2 * 3 + 10 * 4) / 14
eq_(assembled_variant_sequence.mean_coverage(), expected_mean_coverage)
def test_variant_sequence_read_names():
vs = VariantSequence(prefix="A",
alt="C",
suffix="T",
reads=[
AlleleRead(prefix="A",
allele="C",
suffix="T",
name="1"),
AlleleRead(prefix="A",
allele="C",
suffix="T",
name="2")
])
eq_(vs.read_names, {"1", "2"})
def test_assembly_unrelated_sequences():
# 2 overlapping sequences, 1 with a different suffix,
# and 2 totally unrelated sequences
variant_sequences = [
VariantSequence(prefix="CCC", alt="T", suffix="GGG", reads={"1"}),
VariantSequence(prefix="TCCC", alt="T", suffix="G", reads={"2"}),
VariantSequence(prefix="CCC", alt="T", suffix="AAA", reads={"3"}),
VariantSequence(prefix="AGG", alt="T", suffix="CGG", reads={"4"}),
VariantSequence(prefix="CAC", alt="T", suffix="TTT", reads={"5"})
]
results = iterative_overlap_assembly(variant_sequences, min_overlap_size=1)
eq_(len(results), 4)
# first two sequences were overlapping
count_multiple = 0
count_singleton = 0
for result in results:
# all but one result are singletons
if len(result.reads) > 1:
eq_(result.reads, {"1", "2"})
count_multiple += 1
else:
count_singleton += 1
eq_(3, count_singleton)
eq_(1, count_multiple)
def test_variant_sequence_combine():
vs1 = VariantSequence(prefix="A", alt="C", suffix="GG", reads={"1"})
vs2 = VariantSequence(prefix="AA", alt="C", suffix="GG", reads={"2"})
vs_result_1_to_2 = vs1.combine(vs2)
expected = VariantSequence(prefix="AA",
alt="C",
suffix="GG",
reads={"1", "2"})
eq_(vs_result_1_to_2, expected)
# shouldn't matter which sequence is first as an argument to the combine
# function
vs_result_2_to_1 = vs2.combine(vs1)
eq_(vs_result_2_to_1, expected)
def test_variant_sequence_overlaps():
# AAA|GG|TT
vs_3A = VariantSequence(
prefix="AAA",
alt="GG",
suffix="TT",
reads=[AlleleRead(prefix="AAA", allele="GG", suffix="TT", name="1")])
# AA|GG|TT
vs_2A = VariantSequence(
prefix="AA",
alt="GG",
suffix="TT",
reads=[AlleleRead(prefix="AA", allele="GG", suffix="TT", name="1")])
for min_overlap_size in [1, 2, 3, 4, 5, 6]:
assert vs_3A.left_overlaps(vs_2A, min_overlap_size=min_overlap_size), \
"Expected %s to overlap %s from left (min overlap size=%d)" % (
vs_3A, vs_2A, min_overlap_size)
assert not vs_2A.left_overlaps(vs_3A, min_overlap_size=min_overlap_size), \
"Expected %s to not overlap %s from left (min overlap size=%d)" % (
vs_2A, vs_3A, min_overlap_size)
assert not vs_3A.left_overlaps(vs_2A, min_overlap_size=7), \
"Unexpected overlap between %s and %s for min_overlap_size=7" % (
vs_3A, vs_2A)
def make_inputs_for_tp53_201_variant(
cdna_prefix="ATG",
cdna_suffix="AGGAGCCGCAGTCAGAT",
n_bad_nucleotides_at_start=0,
mismatches_before_variant=0,
mismatches_after_variant=14, # the read is that much longer than the reference (17 vs 3)
reference_context_size=3):
"""
Parameters
----------
cdna_prefix : str
Transcript nucleotides before the variant that we're pretending
got detected from RNA-seq reads.
cdna_suffix : str
Transcript nucleotides after the variant that we're pretending
got detected from RNA-seq reads.
n_bad_nucleotides_at_start : int
Number of nucleotides we expect to get trimmed from the
beginning of the variant sequence while matching to a reference context.
mismatches_before_variant : int
Expected number of nucleotide mismatches in the result before
the variant locus.
reference_context_size : int
Number of nucleotides before the variant locus to try matching
against a reference transcript.
"""
# TP53-201 is an isoform of TP53 which seems to lack untranslated
# regions so the sequence is:
# First exon: chr17 7,676,594 - 7,676,521
# ATG|GAG|GAG|CCG|CAG|TCA|GAT...
# -M-|-E-|-E-|-P-|-Q-|-S-|-D-
# we're assuming a variant
# chr17. 7,676,591 C>T which changes GAG (E) > AAG (K)
variant = Variant("chr17", 7676591, "C", "T", "GRCh38")
# TP53-201
transcript = variant.ensembl.transcripts_by_name("TP53-201")[0]
effect = variant.effect_on_transcript(transcript)
eq_(effect.__class__.__name__, "Substitution")
eq_(effect.aa_ref, "E")
eq_(effect.aa_alt, "K")
cdna_alt = "A"
# genomic DNA is the reverse complement of the cDNA
# for TP53-001 since it's on the negative strand
gdna_prefix = reverse_complement_dna(cdna_suffix)
gdna_alt = reverse_complement_dna(cdna_alt)
gdna_suffix = reverse_complement_dna(cdna_prefix)
# variant sequence supported by two reads
# one fully spanning the variant sequence
# and another missing the last nucleotide
fully_overlapping_read = AlleleRead(prefix=gdna_prefix,
allele=gdna_alt,
suffix=gdna_suffix,
name="full-overlap")
# testing the prefix and allele to make sure they have the expected
# TP53-201 sequence but the suffix might change depending on what's
# passed in as cdna_prefix
if cdna_suffix == "AGGAGCCGCAGTCAGAT":
eq_(fully_overlapping_read.prefix, "ATCTGACTGCGGCTCCT")
eq_(fully_overlapping_read.allele, "T")
partially_overlapping_read = AlleleRead(prefix=gdna_prefix,
allele=gdna_alt,
suffix=gdna_suffix[:-1],
name="partial-overlap")
if cdna_suffix == "AGGAGCCGCAGTCAGAT":
eq_(partially_overlapping_read.prefix, "ATCTGACTGCGGCTCCT")
eq_(partially_overlapping_read.allele, "T")
variant_sequence = VariantSequence(
prefix=gdna_prefix,
alt=gdna_alt,
suffix=gdna_suffix,
reads=[fully_overlapping_read, partially_overlapping_read])
assert isinstance(variant_sequence, VariantSequence)
prefix_length = len(cdna_prefix) - n_bad_nucleotides_at_start
reference_coding_sequence_key = ReferenceCodingSequenceKey.from_variant_and_transcript(
variant=variant,
transcript=transcript,
context_size=reference_context_size)
assert isinstance(reference_coding_sequence_key,
ReferenceCodingSequenceKey)
reference_context = ReferenceContext.from_reference_coding_sequence_key(
key=reference_coding_sequence_key,
variant=variant,
transcripts=[transcript])
assert isinstance(reference_context, ReferenceContext)
expected = VariantSequenceInReadingFrame(
cdna_sequence=cdna_prefix[-prefix_length:] + cdna_alt + cdna_suffix,
offset_to_first_complete_codon=prefix_length % 3,
variant_cdna_interval_start=prefix_length,
variant_cdna_interval_end=prefix_length + 1,
reference_cdna_sequence_before_variant="ATG"[-prefix_length:],
reference_cdna_sequence_after_variant=
"AGGAGCCGCAGTCAGAT"[:reference_context_size],
number_mismatches_before_variant=mismatches_before_variant,
number_mismatches_after_variant=mismatches_after_variant)
assert isinstance(expected, VariantSequenceInReadingFrame)
return variant_sequence, reference_context, expected
def test_variant_sequence_contains():
# AA|C|T
vs_longer_prefix = VariantSequence(prefix="AA",
alt="C",
suffix="T",
reads=[
AlleleRead(prefix="AA",
allele="C",
suffix="T",
name="longer_prefix")
])
# A|C|TT
vs_longer_suffix = VariantSequence(prefix="A",
alt="C",
suffix="TT",
reads=[
AlleleRead(prefix="A",
allele="C",
suffix="TT",
name="longer_suffix")
])
# A|C|T
vs_short = VariantSequence(
prefix="A",
alt="C",
suffix="T",
reads=[AlleleRead(prefix="A", allele="C", suffix="T", name="short")])
# two longer sequences contain the shorter subsequence
assert vs_longer_prefix.contains(vs_short), \
"Expected %s to contain %s" % (vs_longer_prefix, vs_short)
assert vs_longer_suffix.contains(vs_short), \
"Expected %s to contain %s" % (vs_longer_suffix, vs_short)
# other pairs do not contain each other
assert not vs_longer_prefix.contains(vs_longer_suffix), \
"Expected %s to not contain %s" % (vs_longer_prefix, vs_longer_suffix)
assert not vs_longer_suffix.contains(vs_longer_prefix), \
"Expected %s to not contain %s" % (vs_longer_suffix, vs_longer_prefix)
assert not vs_short.contains(vs_longer_prefix), \
"Expected %s to not contain %s" % (vs_short, vs_longer_prefix)
assert not vs_short.contains(vs_longer_suffix), \
"Expected %s to not contain %s" % (vs_short, vs_longer_suffix)
# Sequences above has 'C' allele whereas this one has 'G'
# A|G|T
vs_different_allele = VariantSequence(
prefix="A",
alt="G",
suffix="T",
reads=[AlleleRead(prefix="A", allele="G", suffix="T", name="short")])
for vs in [vs_longer_suffix, vs_longer_prefix, vs_short]:
assert not vs.contains(vs_different_allele), \
"Expected %s to not contain %s" % (vs, vs_different_allele)
assert not vs_different_allele.contains(vs), \
"Expected %s to not contain %s" % (vs_different_allele, vs)
def test_variant_sequence_len():
vs = VariantSequence(prefix="AA", alt="C", suffix="TT", reads=[])
eq_(len(vs), 5)
def test_assembly_1_sequence():
vs = VariantSequence(prefix="CCC", alt="T", suffix="GGG", reads={"1"})
eq_(iterative_overlap_assembly([vs]), [vs])
def test_variant_sequence_add_reads():
vs = VariantSequence(prefix="A", alt="C", suffix="G", reads={"1"})
# adding reads '2' and '3', sometimes multiple times
vs_result = vs.add_reads("2").add_reads("1").add_reads("2").add_reads("3")
expected = VariantSequence(prefix="A", alt="C", suffix="G", reads={"1", "2", "3"})
eq_(vs_result, expected)
def test_variant_sequence_contains():
# AA|C|T
vs_longer_prefix = VariantSequence(
prefix="AA",
alt="C",
suffix="T",
reads=[
AlleleRead(
prefix="AA", allele="C", suffix="T", name="longer_prefix")])
# A|C|TT
vs_longer_suffix = VariantSequence(
prefix="A",
alt="C",
suffix="TT",
reads=[
AlleleRead(
prefix="A", allele="C", suffix="TT", name="longer_suffix")])
# A|C|T
vs_short = VariantSequence(
prefix="A",
alt="C",
suffix="T",
reads=[
AlleleRead(
prefix="A", allele="C", suffix="T", name="short")])
# two longer sequences contain the shorter subsequence
assert vs_longer_prefix.contains(vs_short), \
"Expected %s to contain %s" % (vs_longer_prefix, vs_short)
assert vs_longer_suffix.contains(vs_short), \
"Expected %s to contain %s" % (vs_longer_suffix, vs_short)
# other pairs do not contain each other
assert not vs_longer_prefix.contains(vs_longer_suffix), \
"Expected %s to not contain %s" % (vs_longer_prefix, vs_longer_suffix)
assert not vs_longer_suffix.contains(vs_longer_prefix), \
"Expected %s to not contain %s" % (vs_longer_suffix, vs_longer_prefix)
assert not vs_short.contains(vs_longer_prefix), \
"Expected %s to not contain %s" % (vs_short, vs_longer_prefix)
assert not vs_short.contains(vs_longer_suffix), \
"Expected %s to not contain %s" % (vs_short, vs_longer_suffix)
# Sequences above has 'C' allele whereas this one has 'G'
# A|G|T
vs_different_allele = VariantSequence(
prefix="A",
alt="G",
suffix="T",
reads=[
AlleleRead(
prefix="A", allele="G", suffix="T", name="short")])
for vs in [vs_longer_suffix, vs_longer_prefix, vs_short]:
assert not vs.contains(vs_different_allele), \
"Expected %s to not contain %s" % (vs, vs_different_allele)
assert not vs_different_allele.contains(vs), \
"Expected %s to not contain %s" % (vs_different_allele, vs)