# Evaluating Pruning Methods in Gene Network Inference

## TL;DR

This paper evaluates pruning methods in gene network inference, revealing that while pruning can simplify networks, it may also reduce accuracy, highlighting the need for careful strategy selection.

## Contribution

The study systematically assesses ARACNE and Phixer pruning methods, demonstrating potential accuracy trade-offs in gene network inference.

## Key findings

- Pruning may negatively impact inference accuracy.
- Redundant connections like feed forward loops are common.
- Pruning strategies should be chosen carefully based on context.

## Abstract

One challenge in gene network inference is distinguishing between direct and indirect regulation. Some algorithms, including ARACNE and Phixer, approach this problem by using pruning methods to eliminate redundant edges in an attempt to explain the observed data with the simplest possible network structure. However, we hypothesize that there may be a cost in accuracy to simplifying the predicted networks in this way, especially due to the prevalence of redundant connections, such as feed forward loops, in gene networks. In this paper, we evaluate the pruning methods of ARACNE and Phixer, and score their accuracy using receiver operating characteristic curves and precision-recall curves. Our results suggest that while pruning can be useful in some situations, it may have a negative effect on overall accuracy that has not been previously studied. Researchers should be aware of both the advantages and disadvantages of pruning when inferring networks, in order to choose the best inference strategy for their experimental context.