Prediction of Synaptically Localized RNAs in Human Neurons Using Developmental Brain Gene Expression Data

Area of Science:

• Neuroscience
• Computational Biology
• Genomics

Background:

• Synapses are crucial for neuronal communication, with RNA playing key roles in their function.
• Previous research on synaptic RNA pools focused on rodents, leaving human synaptic transcriptomes largely unexplored.
• Studying human synaptic RNA is experimentally challenging, hindering understanding of neuronal processes and diseases.

Purpose of the Study:

• To develop a predictive model for identifying human RNAs localized to synapses.
• To overcome limitations in empirical data for human synaptic RNAs.
• To provide a prioritized list of candidate synaptic RNAs for further experimental validation.

Main Methods:

• A novel machine learning method, PredSynRNA, was developed.
• Training data for dendritically localized RNAs were derived from rodent studies.
• RNA sequence and gene expression data were utilized as predictive features.
• Model performance was evaluated using various learning algorithms and an independent test dataset.

Main Results:

• Machine learning models incorporating developmental brain gene expression features demonstrated superior performance in predicting synaptic RNA localization.
• PredSynRNA successfully predicted and prioritized candidate RNAs localized to human synapses.
• The model's learned expression features were examined and validated.

Conclusions:

• PredSynRNA offers a powerful computational approach to identify human synaptic RNAs.
• The findings provide valuable targets for experimental research on neuronal function and neurological disorders.
• This work advances the understanding of the human synaptic transcriptome and its role in brain health and disease.