Use abPOA for gap alignments

[glennhickey]

This replaces the wfa-based gap sequence aligner with abPOA. The default cactus scoring parameters are hardcoded in. It passes all tests but I haven't yet tested it at scale. In particular it will only work to about ~100kb -- should be fine in practice but at the very least needs an explicit check.

@benedictpaten do you mind testing it builds and runs for you?

As an aside, it wouldn't even pass the tests until I sorted the abpoa input by length -- the problem case is below:

printf ">1\nNNNNNNNNNNNNN\n>2\nNNNNNNNNNNNNNNNNNNN\n>3\nNNNNNNNNNNNNNN\n" > x.fa
taffy/submodules/abPOA/bin/abpoa -r 1 -m 0 x.fa
[main] CMD:  taffy/submodules/abPOA/bin/abpoa -r 1 -m 0 x.fa
>1
NNNNNNNNNNNNN-------
>2
NNNNNNNNNNNNN-NNNNNN
>3
NNNNNNNNNNNNNN------
[abpoa_main] Real time: 0.001 sec; CPU: 0.005 sec; Peak RSS: 0.004 GB.

but sorting works

printf ">2\nNNNNNNNNNNNNNNNNNNN\n>3\nNNNNNNNNNNNNNN\n>1\nNNNNNNNNNNNNN\n" > x.sort.fa
taffy/submodules/abPOA/bin/abpoa -r 1 -m 0 x.sort.fa
[main] CMD:  taffy/submodules/abPOA/bin/abpoa -r 1 -m 0 x.sort.fa
>2
NNNNNNNNNNNNNNNNNNN
>3
NNNNNNNNNNNNNN-----
>1
NNNNNNNNNNNNN------
[abpoa_main] Real time: 0.001 sec; CPU: 0.005 sec; Peak RSS: 0.004 GB.

[benedictpaten]

I fixed the makefile, builds fine on my machine. But the Python build is messed up by the inclusion of abPOA. I'm not quite sure the right way to fix it.

[benedictpaten]

Can you try building the Python part on your machine?:

https://github.com/ComparativeGenomicsToolkit/taffy?tab=readme-ov-file#installing-python-library-from-source

[benedictpaten]

actually, maybe if I remove the merge_adjacent_alignments and have that in a separate header file we can keep a separation from the code in the python library...

[glennhickey]

oof, yeah I hadn't quite gotten to the python build. I'm getting a fatal error: abpoa.h: No such file which I imagine is your problem too? Taking a look now...

[benedictpaten]

Okay.. this all works on my mac (i.e. builds/tests okay)