PON-mt-tRNA: a multifactorial probability-based method for classification of mitochondrial tRNA variations.

Area of Science:

• Genetics
• Molecular Biology
• Bioinformatics

Background:

• Transfer RNAs (tRNAs) are crucial for protein synthesis.
• Human tRNA variations are linked to various clinical conditions.
• All known pathogenic tRNA variations occur in mitochondrial tRNAs (mt-tRNAs).

Purpose of the Study:

• To develop a robust method for classifying pathogenic single nucleotide substitutions in human mt-tRNAs.
• To improve the diagnosis and understanding of mt-tRNA-related diseases.
• To provide a publicly accessible tool for variant analysis.

Main Methods:

• Collected and classified human mt-tRNA variations from multiple evidence sources.
• Developed PON-mt-tRNA, a multifactorial probability-based prediction method.
• Integrated machine learning with evidence-based likelihood ratios (segregation, biochemistry, histochemistry).

Main Results:

• PON-mt-tRNA achieved 1.00 accuracy and 0.99 MCC when using all evidence.
• The machine learning component alone showed 0.69 accuracy and 0.39 MCC without external evidence.
• Analysis revealed variations in loops are more tolerated than stem variations, with the anticodon loop harboring more predicted pathogenic variants.

Conclusions:

• PON-mt-tRNA is a highly accurate tool for classifying pathogenic mt-tRNA variants.
• The tool aids in diagnosing mitochondrial diseases caused by mt-tRNA mutations.
• PON-mt-tRNA is accessible online for research and clinical applications.