The threshold is a percentage, not a ratio.

('segmented.vdj.fa' may be not be the best filename choice, we should change it.)

'removeChevauchement' and possibly other infos

This is a file specific to each people. No reason for versioning

Because of the value for min_cover_representative that has changed (commit
4ee616a) the value computed in that case is now 4 instead of 5. So the
representative is slightly longer which shifted the positions.

The library is now built in core, where we have the list of dependencies
(dep.mk). Before dep.mk was in algo/ but the files it was refering to were in
core/. Thus the dep.mk file was useless.

scatterplot.js : change methods to compute blobs radius, detect if blob represent a clone or a sequence and use correct size to compute radius

scatterplot.js : size analysis , detect if blob represent a clone/cluster or a window/sequence and use correct size

- delete select() function (use closure instead)

- clone mouseover have to work only on clone info and not on cluster info

used to know if we have to display only the representative of the cluster/clone or each windows/sequences

can now do the same for window/sequence

Now that a representative is almost always found, it is better to limit short
examples to a small number of clones.

This enables to have representatives even when setting -R to very small values.
As a side effect, option -A should return representatives for every clone (-R is 1).

Changes in some sequences (in particular IGHV7-40*03) are responsible for
those number changes. There is no such changes when using the previous
germline dataset.

For information IGHV7-40*03 changed from:
>Z12320|IGHV7-40*03|Homo sapiens|P|V-REGION|1..291|289 nt|1| | | | |289+0=289|partial in 5'| |
ctgcagctggtgcagtctgggcctgaggtgaagaagcctggggcctcagtgaaggtctcc
tataagtcttctggttacaccttcaccatctatggtatgaattgggtatgatagacccct
ggacagggctttgagtggatgtgatggatcatcacctacactgggaacccaacgtatacc
cacggcttcacaggatggtttgtcttctccatggacacgtctgtcagcacggcgtgtctt
cagatcagcagcctaaaggctgaggacacggccgagtattactgtgcga
to:
>AC241995|IGHV7-40*03|Homo sapiens|P|V-REGION|10101..10396|296 nt|1| | | | |296+0=296| | |
ttttcaatagaaaagtcaaataatctaagtgtcaatcagtggatgattagataaaatatg
atatatgtaaatcatggaatactatgcagccagtatggtatgaattcagtgtgaccagcc
cctggacaagggcttgagtggatgggatggatcatcacctacactgggaacccaacatat
accaacggcttcacaggacggtttctattctccatggacacctctgtcagcatggcgtat
ctgcagatcagcagcctaaaggctgaggacacggccgtgtatgactgtatgagaga

Maybe we don't need to monitor default behaviour changes in every test.
So I set the seed as a parameter to avoid test failures.

For IGH default seed changed (from #####-##### to ######-######) which changes
segmentation results (and improves them!).

Should we provide the seed to be used as a parameter to avoid such changes? Or
should we let the test as it is to monitor changes in default behaviour?

We have already MAX_CLONES limiting the default output.
In a typical MRD setup, the previous value of 0.1 was not meaningful.

In fact, it could be faster to use MapKmer for very short sets of reads...
but Vidjil focuses on HTS datasets.

- give an html box for the star icon (needed for watir test)

- some class name have changed since
- size displayed can be diffrent from the previous version (due to precision)