CooccurrenceAffinity: An R package for computing a novel metric of affinity in co-occurrence data that corrects for...

Area of Science:

• Ecology
• Bioinformatics
• Computational Biology

Background:

• Traditional co-occurrence indices are flawed, showing sensitivity to prevalence and ambiguity in association strength.
• Previous research identified fundamental issues with common indices and introduced a novel association parameter, alpha, with its maximum likelihood estimate (MLE).

Purpose of the Study:

• Introduce the CooccurrenceAffinity R package for computing the alpha MLE.
• Provide tools for analyzing co-occurrence data from contingency tables and presence-absence matrices.
• Present novel functions for calculating and evaluating median and confidence intervals for the alpha metric.

Main Methods:

• Developed the CooccurrenceAffinity R package implementing alpha MLE calculations.
• Included functions for analyzing 2x2 contingency tables and m x n presence-absence matrices.
• Implemented functions for computing and evaluating median and confidence intervals, including true coverage probability.

Main Results:

• The CooccurrenceAffinity package offers user-friendly, end-to-end analysis for co-occurrence data.
• It computes the alpha MLE, median intervals, and confidence intervals for association.
• The package also provides traditional indices (Jaccard, Sørensen-Dice, Simpson) for comparison.

Conclusions:

• CooccurrenceAffinity provides a robust and efficient solution for co-occurrence data analysis.
• The package enhances ecological and biological association studies by offering a reliable novel metric.
• It facilitates improved understanding of species or entity interactions through advanced statistical methods.