ncov icon indicating copy to clipboard operation
ncov copied to clipboard

Allow users to turn off min and max date filters for frequencies

Open huddlej opened this issue 2 years ago • 0 comments

Context

By default, augur frequencies sets a min and max date based on the collection dates present in the input data. The ncov workflow explicitly hardcodes the --min-date and --max-date arguments for augur frequencies, though, and uses default values of 1 year ago and today for these arguments, respectively.

Users may apply date filters to their input data (e.g., to analyze specific periods of SARS-CoV-2 evolution) that conflict with the default frequency date ranges calculated by the workflow. This means users must duplicate information about date ranges in a separate frequencies config parameter of the workflow. For users with multiple different temporal builds (like Katie's SARS-CoV-2 time window analyses), even this config parameter does not achieve the desired result.

Description

User should be able to turn off default min/max frequencies dates without specifying their own replacement values in the configuration file.

Possible solution

One possible solution that would maintain backward compatibility with current functionality would be to accept a negative boolean value for the min/max dates in the frequencies section of the config file that would disable the passing of the min/max date arguments to augur frequencies. For example, the following configuration section would skip min/max date filtering and allow augur frequencies to set its own range based on the input data:

frequencies:
  min_date: false
  max_date: false

I would argue that users should not have to do anything to get this behavior and that the current default behavior should be opt-in through something like date offsets we've discussed in augur filter. However, it seems too late to make that kind of change without breaking expectations for users who rely on the current defaults.

huddlej avatar Dec 16 '21 20:12 huddlej