IMPARO: inferring microbial interactions through parameter optimisation.

Area of Science:

• Microbiology
• Computational Biology
• Bioinformatics

Background:

• Microbial Interaction Networks (MINs) are crucial for understanding bacterial communities.
• Current methods infer MINs from microbial abundance profiles, often using the Lotka-Volterra model.
• Existing research lacks biologically meaningful mathematical models for MINs and overlooks multiple potential solutions.

Purpose of the Study:

• To present IMPARO, a new method for inferring microbial interactions.
• To incorporate biologically meaningful mathematical models for both abundance profiles and MINs.
• To address the limitation of unique solutions in MIN inference and explore multiple possibilities.

Main Methods:

• IMPARO infers microbial interactions through parameter optimization.
• The method utilizes biologically relevant models for abundance profiles and MINs.
• It assesses the accuracy of reconstructed abundance profiles for different inferred MINs.

Main Results:

• IMPARO can infer multiple MINs with comparable reconstructed abundance profile accuracy.
• The study demonstrates that a unique MIN solution is not always satisfactory or biologically relevant.
• The method successfully identified known gut microbiome interactions, validated by in-vitro experiments.

Conclusions:

• IMPARO effectively infers microbial interactions in human microbiome samples and simulated data.
• The study underscores the significance of considering multiple potential solutions when inferring MINs.
• This approach enhances the biological interpretability of microbial community structures.