def testMultipleQueryAndSubjectGaps(self):
"""
An argument with multiple query and a subject gaps must produce the
expected result.
"""
self.assertEqual((3, 2), countGaps('-GG-34-T-T39F-'))
def normalizeHSP(hsp, queryLen, diamondTask):
"""
Examine an HSP and return information about where the query and subject
match begins and ends. Return a dict with keys that allow the query to
be displayed against the subject. The returned readStartInSubject and
readEndInSubject indices are offsets into the subject. I.e., they
indicate where in the subject the query falls.
In the returned object, all indices are suitable for Python string
slicing etc. We must be careful to convert from the 1-based offsets
found in DIAMOND output properly.
hsp['frame'] is a value from {-3, -2, -1, 1, 2, 3}. The sign indicates
negative or positive sense (i.e., the direction of reading through the
query to get the alignment). The frame value is the nucleotide match offset
modulo 3, plus one (i.e., it tells us which of the 3 possible query reading
frames was used in the match).
NOTE: the returned readStartInSubject value may be negative. We consider
the subject sequence to start at offset 0. So if the query string has
sufficient additional nucleotides before the start of the alignment
match, it may protrude to the left of the subject. Similarly, the returned
readEndInSubject can be greater than the subjectEnd.
@param hsp: an HSP in the form of a C{dict}, built from a DIAMOND record.
All passed offsets are 1-based.
@param queryLen: the length of the query sequence.
@param diamondTask: The C{str} command-line matching algorithm that was
run (either 'blastx' or 'blastp').
@return: A C{dict} with C{str} keys and C{int} offset values. Keys are
readStart
readEnd
readStartInSubject
readEndInSubject
subjectStart
subjectEnd
The returned offset values are all zero-based.
"""
queryGaps, subjectGaps = countGaps(hsp['btop'])
# Make some variables using Python's standard string indexing (start
# offset included, end offset not). No calculations in this function
# are done with the original 1-based HSP variables.
queryStart = hsp['query_start'] - 1
queryEnd = hsp['query_end']
subjectStart = hsp['sbjct_start'] - 1
subjectEnd = hsp['sbjct_end']
queryReversed = hsp['frame'] < 0
# Query offsets must be ascending, unless we're looking at blastx output
# and the query was reversed for the match.
if queryStart >= queryEnd:
if diamondTask == 'blastx' and queryReversed:
# Compute new query start and end indices, based on their
# distance from the end of the string.
#
# Above we took one off the start index, so we need to undo
# that (because the start is actually the end). We didn't take
# one off the end index, and need to do that now (because the
# end is actually the start).
queryStart = queryLen - (queryStart + 1)
queryEnd = queryLen - (queryEnd - 1)
else:
_debugPrint(hsp, queryLen, locals(), 'queryStart >= queryEnd')
if diamondTask == 'blastx':
# In DIAMOND blastx output, subject offsets are based on protein
# sequence length but queries (and the reported offsets) are
# nucleotide. Convert the query offsets to protein because we will
# plot against the subject (protein).
#
# Convert queryLen and the query nucleotide start and end offsets
# to be valid for the query after translation to AAs. When
# translating, DIAMOND may ignore some nucleotides at the start
# and/or the end of the original DNA query. At the start this is
# due to the frame in use, and at the end it is due to always using
# three nucleotides at a time to form codons.
#
# So, for example, a query of 6 nucleotides that is translated in
# frame 2 (i.e., the translation starts from the second nucleotide)
# will have length 1 as an AA sequence. The first nucleotide is
# ignored due to the frame and the last two due to there not being
# enough final nucleotides to make another codon.
#
# In the following, the subtraction accounts for the first form of
# loss and the integer division for the second.
initiallyIgnored = abs(hsp['frame']) - 1
queryLen = (queryLen - initiallyIgnored) // 3
queryStart = (queryStart - initiallyIgnored) // 3
queryEnd = (queryEnd - initiallyIgnored) // 3
# unmatchedQueryLeft is the number of query bases that will extend
# to the left of the start of the subject in our plots.
unmatchedQueryLeft = queryStart
# Set the query offsets into the subject.
queryStartInSubject = subjectStart - unmatchedQueryLeft
queryEndInSubject = queryStartInSubject + queryLen + queryGaps
_sanityCheck(subjectStart, subjectEnd, queryStart, queryEnd,
queryStartInSubject, queryEndInSubject, hsp, queryLen,
subjectGaps, queryGaps, locals())
return {
'readStart': queryStart,
'readEnd': queryEnd,
'readStartInSubject': queryStartInSubject,
'readEndInSubject': queryEndInSubject,
'subjectStart': subjectStart,
'subjectEnd': subjectEnd,
}
def testOneSubjectGap(self):
"""
An argument with just a subject gap must produce the expected result.
"""
self.assertEqual((0, 1), countGaps('G-'))
def testOneQueryAndOneSubjectGap(self):
"""
An argument with a query and a subject gap must produce the expected
result.
"""
self.assertEqual((1, 1), countGaps('G--G'))
def testOneQueryGap(self):
"""
An argument with just a query gap must produce the expected result.
"""
self.assertEqual((1, 0), countGaps('-G'))
def testLettersButNoGaps(self):
"""
An argument with just letters must produce the expected result.
"""
self.assertEqual((0, 0), countGaps('FGAC'))
def testNumberOnly(self):
"""
An argument with just a number must produce the expected result.
"""
self.assertEqual((0, 0), countGaps('88'))
def testEmpty(self):
"""
An argument with an empty string must produce the expected result.
"""
self.assertEqual((0, 0), countGaps(''))
def normalizeHSP(hsp, queryLen, diamondTask):
"""
Examine an HSP and return information about where the query and subject
match begins and ends. Return a dict with keys that allow the query to
be displayed against the subject. The returned readStartInSubject and
readEndInSubject indices are offsets into the subject. I.e., they
indicate where in the subject the query falls.
In the returned object, all indices are suitable for Python string
slicing etc. We must be careful to convert from the 1-based offsets
found in DIAMOND output properly.
hsp['frame'] is a value from {-3, -2, -1, 1, 2, 3}. The sign indicates
negative or positive sense (i.e., the direction of reading through the
query to get the alignment). The frame value is the nucleotide match offset
modulo 3, plus one (i.e., it tells us which of the 3 possible query reading
frames was used in the match).
NOTE: the returned readStartInSubject value may be negative. We consider
the subject sequence to start at offset 0. So if the query string has
sufficient additional nucleotides before the start of the alignment
match, it may protrude to the left of the subject. Similarly, the returned
readEndInSubject can be greater than the subjectEnd.
@param hsp: an HSP in the form of a C{dict}, built from a DIAMOND record.
All passed offsets are 1-based.
@param queryLen: the length of the query sequence.
@param diamondTask: The C{str} command-line matching algorithm that was
run (either 'blastx' or 'blastp').
@return: A C{dict} with C{str} keys and C{int} offset values. Keys are
readStart
readEnd
readStartInSubject
readEndInSubject
subjectStart
subjectEnd
The returned offset values are all zero-based.
"""
queryGaps, subjectGaps = countGaps(hsp['btop'])
# Make some variables using Python's standard string indexing (start
# offset included, end offset not). No calculations in this function
# are done with the original 1-based HSP variables.
queryStart = hsp['query_start'] - 1
queryEnd = hsp['query_end']
subjectStart = hsp['sbjct_start'] - 1
subjectEnd = hsp['sbjct_end']
queryReversed = hsp['frame'] < 0
# Query offsets must be ascending, unless we're looking at blastx output
# and the query was reversed for the match.
if queryStart >= queryEnd:
if diamondTask == 'blastx' and queryReversed:
# Compute new query start and end indices, based on their
# distance from the end of the string.
#
# Above we took one off the start index, so we need to undo
# that (because the start is actually the end). We didn't take
# one off the end index, and need to do that now (because the
# end is actually the start).
queryStart = queryLen - (queryStart + 1)
queryEnd = queryLen - (queryEnd - 1)
else:
_debugPrint(hsp, queryLen, locals(), 'queryStart >= queryEnd')
if diamondTask == 'blastx':
# In DIAMOND blastx output, subject offsets are based on protein
# sequence length but queries (and the reported offsets) are
# nucleotide. Convert the query offsets to protein because we will
# plot against the subject (protein).
#
# Convert queryLen and the query nucleotide start and end offsets
# to be valid for the query after translation to AAs. When
# translating, DIAMOND may ignore some nucleotides at the start
# and/or the end of the original DNA query. At the start this is
# due to the frame in use, and at the end it is due to always using
# three nucleotides at a time to form codons.
#
# So, for example, a query of 6 nucleotides that is translated in
# frame 2 (i.e., the translation starts from the second nucleotide)
# will have length 1 as an AA sequence. The first nucleotide is
# ignored due to the frame and the last two due to there not being
# enough final nucleotides to make another codon.
#
# In the following, the subtraction accounts for the first form of
# loss and the integer division for the second.
initiallyIgnored = abs(hsp['frame']) - 1
queryLen = (queryLen - initiallyIgnored) // 3
queryStart = (queryStart - initiallyIgnored) // 3
queryEnd = (queryEnd - initiallyIgnored) // 3
# unmatchedQueryLeft is the number of query bases that will extend
# to the left of the start of the subject in our plots.
unmatchedQueryLeft = queryStart
# Set the query offsets into the subject.
queryStartInSubject = subjectStart - unmatchedQueryLeft
queryEndInSubject = queryStartInSubject + queryLen + queryGaps
_sanityCheck(subjectStart, subjectEnd, queryStart, queryEnd,
queryStartInSubject, queryEndInSubject, hsp, queryLen,
subjectGaps, queryGaps, locals())
return {
'readStart': queryStart,
'readEnd': queryEnd,
'readStartInSubject': queryStartInSubject,
'readEndInSubject': queryEndInSubject,
'subjectStart': subjectStart,
'subjectEnd': subjectEnd,
}
