An accurate and interpretable model for siRNA efficacy prediction.

Area of Science:

• Molecular Biology
• Bioinformatics
• Genetics

Background:

• Small interfering RNAs (siRNAs) are key tools for gene silencing in research and drug discovery.
• Designing potent siRNAs remains a significant challenge despite advances in RNA interference (RNAi) pathway understanding.

Purpose of the Study:

• To develop a simple, interpretable linear model for predicting siRNA efficacy.
• To identify sequence-based features that contribute to siRNA potency.

Main Methods:

• A linear model was developed using basic features of siRNA sequences.
• The model was trained and validated on a large dataset of siRNA sequences.
• Analysis focused on nucleotide preferences at specific positions and the presence of asymmetric sequence motifs.

Main Results:

• The linear model achieved prediction accuracy comparable to complex, state-of-the-art models.
• Nucleotide preferences at specific positions were identified and quantified.
• Short asymmetric motifs within siRNA sequences were found to be highly informative for potency prediction, sometimes as much as positional nucleotide preferences.

Conclusions:

• A simple, interpretable linear model can accurately predict siRNA potency.
• The model provides biological insights into sequence features driving siRNA efficacy.
• The prediction tool is publicly available online for researchers.