Add filter option that restricts primer design to regions specific to a target class

[widdowquinn]

In order to maximise the ability to design primers belonging to a single target class, we propose to use the pdp filter subcommand to restrict the primer design region for a single target class in the config file, with a command like:

pdp filter --target CLASS

Where CLASS is a single target class defined in the configuration file for the run.

With this option specified, an all-vs-all mummer comparison of input genomes will be performed, and genomes divided into two groups: those having CLASS specified (targets), and those not having CLASS specified (non-targets).

For those genomes with the CLASS property, the mummer alignments against non-CLASS genomes will be identified, and any regions contributing to such an alignment will be excluded by the filter.

This process could be extended with the command

pdp filter --target_specific CLASS

which would carry out the same alignments, but further inspect the alignments of CLASS genomes to each other and retain only those regions that are common to all members of CLASS.

[widdowquinn]

As an implementation, a first thought is:

• all-vs-all genome comparisons with mummer
• for each genome with CLASS
  • collect alignments against non-CLASS genomes as intervals on the query genome
  • identify the union of aligned regions on the query genome
  • generate an exclusion .bed file that is the inverse of the union of aligned regions

The .bed file will then define the region in each CLASS genome to which primer sets may be designed, when using the pdp eprimer3 --filter subcommand.