Algorithm uses a lot of RAM.

In production we observe this algorithm using 3.52 GB to process ~35 sequences.

This caused the production server to OOM Kill the process.

Please investigate memory optimizations.