GitLab

Makes it more robust to changes in make utility

Now all should-vdj tests are also run on the rev-comped sequences.
This increases the test time, but improves our robustness.

The 4 sequences added in 6503f6ae now give correct results.
All the 68 (now 72) test sequences are now passing on both strands.

68 sequences are currently passing tests.
Rev-comping these sequences ('python should-vdj-to-tap.py -r'), the tests
fail on these 4 sequences.

Differs on + and - strand. Incorrect on both

Since the 2015.07 release, many changes improved the V(D)J designation (core/segment.cpp,
in particular c5c781ec, 77af56b8, 185c3bd9, 60a15629), and this is now better tested
(.should-vdj.fa tests, with new curated sequences and an improved test mechanism).

This was a bug hidden since the first release of Vidjil.
It is now corrected, and the test added in 002adbd2 is now passing.

This test was not passing 10 days ago, before the work on close V-J xxx

When entering mnaually curated sequences, this is faster than launching
a full 'make shouldlocus_and_vdj' (which further recompiles everything).

add 6 errors :
 - 2 double DD
 - 3 ambigous
 - 1 real error (the TRB error)

Handle '[xxx]' locus information.

Completes a36f0231.

The actual command line is stored in the .log file.

Tests marked with 'TODO' are expected to fail.
Such tests do not trigger an error exit code,
and the 'TODO' keyword is output in the corresponding line of the .tap file.

After correction, only 3 errors stay :
 - 2 DD2-DD3 not seen
 - 1 ambigous case

.should-vdj tests can now be given with spaces instead of underscores.

It is better to segment the first sequence without D.