NFB-Checker: An AI/ML-Powered Microorganism Nitrogen Fixation Susceptibility Prediction from Gene Collection

Nitrogen fixation is a crucial process involved in many aspects of the ecological life cycle. However, few organismsparticularly kingdom bacteria-are known to fix N<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> in a form that could be used in agribusiness. The ability to identify an organism as an N<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> fixer is a novel case for more identification and application. This study employs functional genes and an XGBoost machine learning model to predict an organism's ability to fix nitrogen, focusing on key orthologs like nifH, nifD, and nifK, essential in the nitrogenase enzyme complex. The model, trained on a dataset of 923 N<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> fixers and 981 non-fixers, achieved an accuracy of 92.6% and an F1 score of 0.92. This research highlights the potential of machine learning in identifying genetic markers for N<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> fixation, providing an efficient alternative to traditional methods, and offers new insights for ecological and agricultural research. The inclusion of specific orthologs enhances the model's predictive accuracy, demonstrating the importance of targeted genetic markers in computational biology.