Resume Multiple Sequence Alignments

[nzmacalasdair]

This adds the ability to resume MSAs which hit resource/runtime limits with panaroo-msa --resume, for both non-codon and codon alignments.

Does this by writing a small JSON with the alignment parameters, which is checked for when an alignment starts. Alignment module then looks for completed output in aligned_gene_sequences (or also aligned_protein_sequences for codon alignments) and skips things which are already there.

Done testing on two small datasets so far, seems to run without issues and produces 100% identical output. May be worth adding an experimental flag in the initial release however.

[gtonkinhill]

Hi Neil,
This is currently hitting an error as the generate_pan_genome_alignment has been changed but not all the references to it have been updated.
Is there a reason it needs to have a 'core' parameter as all genes should be aligned when the 'pan' is specified by the user

[nzmacalasdair]

This is because the checkpoint file (which needs the threshold to be written so the resume can use the same threshold) is currently created within either generate_core_genome_alignment or generate_pan_genome_alignment.

There's no real need for it to be written within those functions though, it's inelegant and I will change that.

Are you happy with the helper function spam in generate_alignments.py? I could clean that up as well.

[gtonkinhill]

Thanks for that. Yes, happy with the spam 😄