Microbiome data analysis via machine learning models: Exploring vital players to optimize kitchen waste composting...

Area of Science:

• Microbiology
• Environmental Science
• Data Science

Background:

• Composting is a vital biological treatment for kitchen waste, relying on microbial communities.
• Precisely identifying the roles of specific microorganisms in composting is challenging using traditional experimental methods alone.

Purpose of the Study:

• To utilize machine learning models for identifying key microbial genera involved in composting.
• To optimize composting systems through data-driven insights into microbial functions.

Main Methods:

• A novel microbiome preprocessing approach was developed.
• Stacking machine learning models were constructed, achieving an R² of approximately 0.8.
• The SHAP (SHapley Additive exPlanations) method was employed to determine the significance of microbial genera.

Main Results:

• The SHAP analysis identified Bacillus, Acinetobacter, Thermobacillus, Pseudomonas, Psychrobacter, and Thermobifida as prominent microbial genera (Shapley values: 3.84–1.24).
• Microbial agents targeting these key genera were prepared and tested.
• Composting systems treated with these agents showed improved quality scores (76.06) compared to controls (70.96).

Conclusions:

• This study successfully employed machine learning to pinpoint crucial microbial players in composting.
• The findings demonstrate that targeting identified key genera can enhance waste decomposition and compost quality.
• This research introduces an innovative strategy for optimizing biological waste treatment processes through microbial insights.