Memory issues

[silenus092]

Hi

I have a memory problem. For example, When I use the Mpoxvirus reference sequence to align with a partial genome or specific coding sequence. The program used 62 GB+ of RAM, and the OS killed the process before it finished the alignment.

I've come across https://github.com/smarco/BiWFA-paper which seems to offer a solution with lower memory usage. Additionally, I've noticed the presence of low-memory modes in the code (WFA_wrap.pxd).

Will you plan to integrate these features in the Python interface pyfwa to enable the BiWFA algorithm or select memory strategies, or is there an alternative way? Please feel free to suggest me.

I also changed the heuristic functions (adaptive and Xdrop ), but they didn't help in my case.

[kcleal]

Hi @silenus092,
It would be nice to add BiWFA, for sure. However, I don't have much time to add this. Perhaps in a month or two I might take a look.

[JosephLalli]

I've just implemented this option in my fork here: https://github.com/josephlalli/pywfa

Very experimental, no idea how exposing this option affects accuracy or introduces other complications.

To use, when initializing your WavefrontAligner, specify memory_mode 'biwfa'.

from pywfa import WavefrontAligner

a = WavefrontAligner(memory_mode='biwfa')