compare_biom.py – Compare the accuracy of biom files (expected and observed) either by observations (default) or by samples.¶
Description:
Usage: compare_biom.py [options]
Input Arguments:
Note
[REQUIRED]
- -e, --exp_trait_table_fp
- The expected trait table (biom format)
- -o, --output_fp
- The output file
[OPTIONAL]
- -c, --compare_observations
- Calculate accuracy values by comparing between observations (instead of between samples) [default: False]
- -n, --normalize
- Convert both expected and observed tables to relative abundances (instead of observations) [default: False]
- -l, --limit_to_expected_observations
- Ignore observations that are not in the expected table[default: False]
- --limit_to_observed_observations
- Ignore observations that are not in the observed table[default: False]
- -s, --shuffle_samples
- Shuffle samples ids randomly before measuring accuracy[default: False]
- --not_relative_abundance_scores
- Round numbers (instead of taking ceil() which is used for RA) before calculating TP,FP,FN,TN [default: False]
Output:
Outputs will be tab delimited file with various accuracy metrics.
Example 1:
Compare an observed table to an expected table using relative abundance
compare_biom.py -e expected_ra.biom -o compare_results_ra.tab observed_ra.biom
Example 2:
Compare an observed table to an expected table using real counts
compare_biom.py --not_relative_abundance -e expected.biom -o compare_results.tab observed.biom