This is quadratic, both in time and in space in the .vidjil output.

Mathieu Giraud authored Aug 02, 2016 and Vidjil Team committed Aug 02, 2016

The Vidjil Morning failed sometimes.

This is our user manual. We do not mention 'browser' anywhere in this documentation except while referring to web browsers.

Some information was scattered in the previous organization.
This file is now the central place to store *developer* information.
We may one day re-split this documentation into several files.

Some external tools would like to provide sequence features with textual information.
Discussion with @flothoni and @duez.

'stop' positions are included in the highlight (5e21e0).

Checked on Demo-X5 both on 2016.03 and 2016.07 outputs (that are both 1-based),
and both on 'cdr3' (computed with -3) and 'id' fields.

This should be an old file.

Now the tests added in the previous commit are working.

Using the same object is not good (at least for tests) since we shift
his position towards 0-based.

All fields go now through a 1-based -> 0-based transformation.
Howvever, we now do the reverse transformation for 5/4/3 fields
(old 5/4/3 fields were 0-based).

This will allow somewhat large sets of reads.

A @rchikhi set with ~150M reads tested with all locus gave an e-value multiplier >2^31,
leading to integer overflow and many spurious clones.

Moreover, the previous multiplier was only an 'int', depending on the actual implementation.
The multiplier is now a float (double) value.

Thanks for @rchikhi for pointing this out.

The other files are generated through a 'make' in tools/tests

Mathieu Giraud authored Jul 13, 2016 and Vidjil Team committed Jul 13, 2016

We want to be able to retrieve the last fuses on every sample set

Follows 7b955792

Vidjil Team authored Jul 13, 2016 and Mikaël Salson committed Jul 13, 2016

English and French versions