GitLab

Note that the check will work on Python 2.6, but not on Python 2.4.

The distribution information is put in the reads entry.
We then have one entry per distribution (under a distribution entry)

Thresholds are not anymore in the number of absolute reads
but rather in the ratio of segmented reads.

Previously, load_clntab() was launched when there was an exception raised by load_vidjil().
This was error prone, moreover we need the get the proper exception when there is a real error in load_vidjil().

Now the check is on the file name ('.clntab'). There could be better strategies, but
the detection of the file type has to be separated from the parsing itself.

The previous test (merging 3 files) was not working.

'sequence.JUNCTION.raw nt seq' is the JUNCTION, including V-Cys and J-Phe/J-Trp,
whereas the CDR3 does not include these amino acids.

The top-level fields are created once when initializing ListWindows(),
whereas sample/ fields are created when loading data into ListWindows()

The computer science department of the Université de Lille is is now CRIStAL

This was introduced by b9cd02c4, but removed with b9cd02c4. This prevents fuse.py from halting
if unknown non-list fields are present.

I am not sure that this is useful (and here it's somewhat different: such values are ignored).
We need more unit and functional tests on fuse.py !

0ee4e4e9 forgot this one

Multiple germlines are correctly handled by the regular fuse.py.

replace .vidjil with .analysis only for extension