GitLab

> pear.py > don't produce output file on failure
> don't reschedule run if pre-process has failed

before > true/false
after > WAIT/QUEUD/RUN/DONE/FAILED

Preprocess filename was an id followed by .fastq. Instead we try to take the
best among the two filenames (or only the first filename, if there is no
second filename). This will be more meaningful to the user and the file
is not necessarily in .fastq format.

The end user should have access to the pre-process log.

There may be pre-processes running on only one file.

Added a parsing of the mixcr assembly report to extract the total number
of reads and add it to the output

This reverts commit c97ab0c531ede8649459168a6cc488a842789d51.
Responsibility for version output has been deferred to the mixcr native
program

Moved some code into fill_field

-t 1 has another meaning for exportClones than for the other
 commands (it is related to stop codons, see:
 https://mixcr.readthedocs.org/en/latest/export.html#command-line-parameters)

Put the results_file data_file stream back in because it is needed

Added run task for mixcr program

task.py : fix run_vidjil > (re)compute the fuse file for every sample_set related to the sequence file

controllers/ default.py results_file.py models/task.py changed signature of schedule_run to accept sample_set_id

Due to the changes to the database model it became necessary to pass
this id in the message signature. As a result, the run_all method needed
to be changed. We decided to limit the scope of run_all for obvious
reasons. It is now run_all_patients