After working with the config in the input-dir to write the application.properties, which is already configured prior to the run, I agree that it makes sense to move any tool-specific configurations, that are needed to be referenced in the runner, into this file under a tool_specific_configurations field.
However, I would like to propose that there are two additional fields added:
tool: exomiser
tool_version: 13.2.0
phenotype_only: True
post_processing:
score_name: combinedScore
sort_order: DESCENDING # or can be ASCENDING
tool_specific_configuration_options:
This is particularly for tools that output several different scores and perhaps want to specify on which score you want to benchmark. The score order is required for the post-processing, i.e., if the best score is considered those with the lowest value (p-value) these will be ranked first in ascending order, or if the larger values are the best scores, then ranked in descending order. This can be added to the tool_specific_configuration_options field, however, it is information required currently in the post-processing methods so would propose that these are in the main part of the config.
Alternatively, if the tool only outputs a single score, like Phen2Gene they can choose to ignore these fields and just pass the arguments for the post-processing in the runner implementation.
After working with the config in the
input-dir
to write theapplication.properties
, which is already configured prior to the run, I agree that it makes sense to move any tool-specific configurations, that are needed to be referenced in the runner, into this file under atool_specific_configurations
field.Currently, the input-dir config looks like so:
However, I would like to propose that there are two additional fields added:
This is particularly for tools that output several different scores and perhaps want to specify on which score you want to benchmark. The score order is required for the post-processing, i.e., if the best score is considered those with the lowest value (p-value) these will be ranked first in ascending order, or if the larger values are the best scores, then ranked in descending order. This can be added to the
tool_specific_configuration_options
field, however, it is information required currently in the post-processing methods so would propose that these are in the main part of the config.Alternatively, if the tool only outputs a single score, like Phen2Gene they can choose to ignore these fields and just pass the arguments for the post-processing in the runner implementation.
@matentzn what do you think?