SPLANG-a synthetic poisson-lognormal-based abundance and network generative model for microbial interaction inference...

Area of Science:

• Microbial ecology and bioinformatics.
• Computational biology and network science.

Background:

• Microbial interactions profoundly influence ecosystems, impacting health and agriculture.
• Inferring microbial interaction networks from abundance data is challenging due to algorithmic validation difficulties.

Purpose of the Study:

• To introduce a generative model for synthesizing microbial interaction networks and abundance data.
• To serve as a test bed for evaluating existing and novel network inference algorithms.

Main Methods:

• Development of a generative model to create synthetic microbial interaction networks and corresponding abundance data.
• Testing four established network inference algorithms using the synthetic data.
• Evaluation of an oracle algorithm combining multiple inference methods.

Main Results:

• Existing network inference algorithms show inadequate accuracy for non-commensalistic (mutualistic or competitive) species interactions.
• The oracle algorithm demonstrated improved predictability but struggled with densely interacting microbial networks.
• Significant limitations exist in current algorithms for accurately reconstructing complex microbial ecologies.

Conclusions:

• Current algorithms require further development and validation for reliable microbial network inference.
• The generative model provides a valuable tool for assessing algorithm performance.
• Accurate reconstruction of microbial interaction networks remains an open challenge.