Memory for kraken2

[wm75]

The memory requirements of this tool are largely dependent on the DB size you're using with it.
For DB sizes see:
https://benlangmead.github.io/aws-indexes/k2
As you can see, there are dramatic size differences.

Can this information be used somehow to guide the memory assigned to the tool?

[bgruening]

If you provide a database=memory mapping we could use that I think.

[natefoo]

I am using DB size + 24 GB:

  toolshed.g2.bx.psu.edu/repos/iuc/kraken2/kraken2/.*:
    cores: 16
    # TODO: make a helper
    mem: |
      import os
      mem = 64
      table_name = "kraken2_databases"
      lookup_column = "value"
      value_column = "path"
      value_template = "{value}/hash.k2d"
      options = job.get_param_values(app)
      lookup_value = options["kraken2_database"]
      table_value = app.tool_data_tables.get(table_name).get_entry(lookup_column, lookup_value, value_column)
      if table_value is not None:
          table_value = value_template.format(value=table_value)
          try:
              mem = os.path.getsize(table_value)/1024**3
              log.debug("Data table '%s' lookup '%s=%s: %s=%s': %s GB",
                        table_name, lookup_column, lookup_value, value_column, table_value, mem)
          except OSError:
              log.exception("Failed to get size of: %s", table_value)
      else:
          log.warning("Data table '%s' lookup '%s=%s: %s=None' returned None!, defaulting to %s",
                      table_name, lookup_column, lookup_value, value_column, mem)
      mem + 24

[nuwang]

This looks like a good rule to add to the TPV db itself? Would you be able to make a PR?

[natefoo]

Yes, I meant to add a helper at some point, just haven't gotten to it yet - I will do that.