Coronavirus samples run out of memory

[donkirkby]

I ran several coronavirus samples under Kive, and many got killed for using too much memory. Looking in the sandbox, it seemed to happen during aln2counts, so I reran that step on my workstation, under slurm with the same memory limit. It also got killed.

I added some debugging output, and I found that it's happening when I use Gotoh to find the location of the repeated base in the consensus sequence. Aligning two full-length SARS-CoV-2 sequences is too much for Gotoh to handle. It succeeds on my workstation outside of Kive, but Slurm notices the huge memory usage and kills it when Kive runs it.

Searching for a small portion of SARS-CoV-2 around the repeated base should be much more efficient.