Prediction of Neurodevelopmental Disorders Based on De Novo Coding Variation

Area of Science:

• Genetics
• Computational Biology
• Neuroscience

Background:

• Early detection of neurodevelopmental disorders (NDDs) significantly improves patient outcomes.
• De novo mutations, particularly coding variations, are implicated in NDDs.
• Identifying individuals with NDD phenotypes using genetic data is a key challenge.

Purpose of the Study:

• To develop an accurate predictive approach for NDDs using de novo coding variation.
• To achieve a very low false positive rate (FPR) in NDD prediction.
• To prioritize NDD risk genes for further research.

Main Methods:

• Utilized a shallow neural network for NDD prediction.
• Integrated de novo likely gene-disruptive and missense variants.
• Incorporated measures of gene constraint and evolutionary conservation.

Main Results:

• Developed a method for accurate NDD prediction with a very low FPR.
• Successfully identified a subset of NDD cases using de novo coding variation.
• Prioritized potential NDD risk genes.

Conclusions:

• De novo coding variation is a valuable marker for predicting NDDs.
• The developed neural network approach offers high accuracy and low FPR.
• This method aids in identifying individuals and genes relevant to NDDs.