some bulk sequence loading tests that nail down current ACGTN behavior. #1633
Conversation
Codecov Report
@@ Coverage Diff @@
## master #1633 +/- ##
=========================================
+ Coverage 69.82% 69.93% +0.1%
=========================================
Files 66 66
Lines 8974 8976 +2
Branches 3060 3062 +2
=========================================
+ Hits 6266 6277 +11
+ Misses 1025 1018 -7
+ Partials 1683 1681 -2
Continue to review full report at Codecov.
|
|
Wow, these tests are like a showcase of horrible inconsistency in loading sequences. |
tests/test_read_parsers.py
Outdated
| kmer = "caggcgcccaccacc".upper() | ||
| assert x.get(kmer) == 1 | ||
|
|
||
| # the 2nd read with this k-mer in it has an N in it; 'consume' will ignore. |
There was a problem hiding this comment.
I'd rephrase the comment. At the moment I am very puzzled after reading it (ah its being ignored so count should be one, but it is asserted to be two ...??)
There was a problem hiding this comment.
Maybe "consume will ignore the invalid base and continue consuming the read, so this kmer after the N should have abundance 2"
|
I'd move these tests to a new file maybe Because it is easy to do I would parametrise the tests (that make sense) on the class so we test all combinations of Count/Node and table/graph. In which case we should get the parametrisation stuff from |
tests/test_read_parsers.py
Outdated
|
|
||
| x.output_partitions(infile, savepath) | ||
|
|
||
| read_names = [ read.name for read in ReadParser(savepath) ] |
There was a problem hiding this comment.
The pep8 🚓 doesn't like the extra white space (hence the travis failure)
|
+1 on merging this before 2.1, then #1590 can update the behaviour and fix these tests (remove the "in the future ..." comments) |
|
Tests moved to |
|
Linux build fails because of a pep8 violation. The OSX build fails because some tests fail and then we exit ungracefully because there are too many open files. Will take a look at the latter. |
|
Ha, locally on OSX all tests pass as well and no "too many open files" warnings. |
With the fixture we can explicitly close all the ReadParser which might help with the too many open files error on OSX Travis
|
Switched to using a fixture for |
|
🎉 |
|
This is now ready for review & merge (although #1661 should be merged first :) |
|
(Upon merge, we should switch #1590 over to be against master.) |
tests/test_sequence_validation.py
Outdated
| savepath = utils.get_temp_filename('foo') | ||
|
|
||
| # read this in using "approved good" behavior w/cleaned_seq | ||
| x = _Nodegraph(8, PRIMES_1m) |
tests/test_sequence_validation.py
Outdated
| return request.param | ||
|
|
||
|
|
||
| @pytest.fixture |
There was a problem hiding this comment.
Now that I've looked this up: can we change this back to yield and change the decorator to @pytest.yield_fixture? Link to the pytest 2.9 docs: http://doc.pytest.org/en/2.9.2/fixture.html#fixture-finalization-executing-teardown-code
|
done
|
|
all comments addressed in e419a7b
|
|
🎉 |
These tests nail down behavior prior to #1590, which, when merged, will alter how we handle non-ACTG characters. Note, no behavior is changed in this PR; it's just (lots of) new tests.
This explicitly puts in place tests for sequences that contain one of lowercase, Ns, and non-ACGTN characters, for:
consume_fastaand all other bulk-sequence loading functions on Hashtables and derived classes;trim_on_abundance,trim_below_abundanceandfind_spectral_error_positionsThis PR includes #1661.
Adds new test file
tests/test_sequence_validation.pyand data filetests/test-data/valid-read-testing.fq.make testDid it pass the tests?make clean diff-coverIf it introduces new functionality inscripts/is it tested?make format diff_pylint_report cppcheck doc pydocstyleIs it wellformatted?
additions are allowed without a major version increment. Changing file
formats also requires a major version number increment.
documented in
CHANGELOG.md? See keepachangelogfor more details.
changes were made?
tested for streaming IO?)