Quantitative Missense Variant Effect Prediction Using Large-Scale Mutagenesis Data.

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Quantitative data on mutation effects on protein function is growing.
• Predicting missense variant effects is crucial for understanding genetic diseases.

Purpose of the Study:

• To develop Envision, a novel predictor for the magnitude of missense variant molecular effects.
• To leverage large-scale experimental mutagenesis datasets as a training resource.

Main Methods:

• Developed Envision using a supervised, stochastic gradient boosting learning algorithm.
• Integrated 21,026 variant effect measurements from nine large-scale mutagenesis datasets.
• Validated Envision on an independent dataset of 2,312 TP53 variants.

Main Results:

• Envision outperforms existing missense variant effect predictors on both large-scale and independent test datasets.
• Envision exhibits more consistent prediction accuracy across amino acids compared to other methods.
• Prediction accuracy improves with the incorporation of additional large-scale mutagenesis data.

Conclusions:

• Envision represents a significant advancement in predicting missense variant effects.
• The tool's performance enhances as more experimental data becomes available.
• Envision predictions are available for proteomes of multiple species.