Dev

[bricoletc]

This PR has two main goals:

• Addressing #17
• Adding more tests and refactoring some of the codebase

Added

• IntervalPartitioner object and associated classes, in interval_partitioner.py This object replaces the partition_alignment_into_intervals function, which was harder to understand, also makes it more easily to modify in order to address #17 and in future. Also added more unit tests for interval partitioning.
• This object, when partitioning the consensus string, now spots if a non_match interval induces any empty sequence, and pads any leading match sequence to it. Unit tests to validate the behaviour, and I ran the following toy example:

  >seq1
  AATAATAATAAATTTTGTATAAACTTTACCCTAG
  >seq2
  AATAATAAT----------------TTACCCTAG
  >seq3
  AATAATAATAAATTTTATATAAACTTTACCCTAG

  Before this PR, results in: AATAATAAT 5 AAATTTTGTATAAACT 6 AAATTTTATATAAACT 6 5 TTACCCTAG (see the empty third allele`

After this PR, results in : AATAATAA 5 TAAATTTTGTATAAACT 6 TAAATTTTATATAAACT 6 T 5 TTACCCTAG ( the T got prepended) (max_nesting 5, min_match_len 7)

This does not mean #17 is fully resolved, i'll now run on real datasets

Modified

• An alignment column of all '-' is no longer counted as a match. I cannot find a good reason for it to be but may be missing something?

• Moved subcommand parsing to subcommands/ dir, results in less cluttered main()

• get_consensus function shorter and easier to understand + add unit tests for it

• refactored interval partition checks (checking each position of alignment is in one and one only interval, and switching non-match to match intervals where needed) into own functions and unit tested them.

• -v for verbose switch added at top-level command line

• Overall I strove to make the main module make_prg_from_msa smaller and terser, for easier reading/modifying later: went from 499 to 385 lines.

[bricoletc]

Hi @mbhall88 @rmcolq , I think this should be merged to master. I can certify that it worked for my gramtools paper analysis, producing PRGs where I found, genotyping on them with gramtools, good concordance with PacBio truth assemblies on two different datasets (TB, Pfalciparum)