conditional_uncovered_probability.py – Calculate the conditional uncovered probability on each sample in an otu table.

Description:

This script calculates the conditional uncovered probability for each sample in an OTU table. It uses the methods introduced in Lladser, Gouet, and Reeder, “Extrapolation of Urn Models via Poissonization: Accurate Measurements of the Microbial Unknown” PLoS 2011.

Specifically, it computes a point estimate and a confidence interval using two different methods. Thus it can happen that the PE is actually outside of the CI.

We only provide the ability to generate 95% (alpha=0.95) CIs. The CIs are ULCL CIs; they provide an upper and lower bound, where the lower bound is conservative. The CIs are constructed using an upper-to-lower bound ratio of 10.

The CI method requires precomputed constants that depend on the lookahead. We only provide constants for r=3..25,30,40,50.

Usage: conditional_uncovered_probability.py [options]

Input Arguments:

Note

[OPTIONAL]

-i, --input_path

Input OTU table filepath. [default: None]

-o, --output_path

Output filepath to store the predictions. [default: None]

-r, --look_ahead

Number of unobserved, new colors necessary for prediction. [default: 25]

-m, --metrics

CUP metric(s) to use. A comma-separated list should be provided when multiple metrics are specified. [default: lladser_pe,lladser_ci]

-s, --show_metrics

Show the available CUP metrics and exit.

Output:

The resulting file(s) is a tab-delimited text file, where the columns correspond to estimates of the cond. uncovered probability and the rows correspond to samples. The output file is compatible with the alpha_diversity output files and thus could be tied into the rarefaction workflow.

Example Output:

	PE	Lower Bound	Upper Bound
PC.354	0.111	0.0245	0.245
PC.124	0.001	0.000564	0.00564

Default case:

To calculate the cond. uncovered probability with the default values, you can use the following command:

conditional_uncovered_probability.py -i otu_table.biom -o cup.txt

Change lookahead:

To change the accuracy of the prediction change the lookahead value. Larger values of r lead to more precise predictions, but might be unfeasable for small samples. For deeply sequenced samples, try increasing r to 50:

conditional_uncovered_probability.py -i otu_table.biom -o cup_r50.txt -r 50