GitLab

Giving two KmerAffects, which one is the more on the left ?

The special germline PSEUDO_GERMLINE_MAX12 uses the usual KmerSegmenter,
but launches computeSegmentation() on the two most frequent KmerAffects of the read.

The 'strand' detection is also more flexible here and purely indicative, as
we could even segment chimera reads.

If you really want a Germline without germlines, please do not try to FineSegment.

Who wants to define Germlines without germlines ?

cleancoverage removes coverage-related files
cleantests removes test-related files
cleanall removes everything.

KmerMultiSegmenter: TOO_FEW_ZERO (instead of UNSEG_NOISY) iff the e_value is bad globally not either on left or on right

That avoids to recompute them each time the getLeftRightProbabilityAtLeastOrAbove() function is called.

Fix 34068e0b, thanks to Valgrind

TRDD2*01 is exactly 9 bp long. With SEED_9, we have in perfect Dd2 recombinations *one* matching k-mer.
The Dd2 TRD+ indexes have a load of around 0.05% and 0.16% (due to Dd3 or J), and thus, with the current e-value
computation, there is no chance to get a left e-value much smaller than 1e-3.

Some tests do not pass anymore with the -e 1e-6 threshold applied on both sides.

Store and check getLeftRightProbabilityAtLeastOrAbove()

core/affectanalyser.{h, cpp}: getLeftRightProbability, p-value at both sides of the segmentation point

We want to be able to test coverage on different rules, without
deleting previous coverage files.

We want to display this thing in the browser.

The test was not working since we were not outputting the file created by fuse
and the right parameters were not passed to format_json.py

Fix 05b02185, thanks to @mikael-s for the bisect.

This doesn't prevent from testing unsegmented sequences
as the output is not UNSEG UNSEG (UNSEG will appear only once).

Conflicts:
	server/web2py/applications/vidjil/controllers/patient.py

remove spurious <div>

Partially revert 1fcc4967, add reword some entries.