Study coordinator: Pr Valérie MALAN, Dr Antonio RAUSELL
Neurodevelopmental disorders affect approximately 2–3% of the general population and present with a wide range of clinical manifestations. Despite their significant collective prevalence, a diagnosis is reached in only about half of the cases. To date, over 1,200 genes have been associated with these clinical conditions.
Genome sequencing technologies have greatly facilitated the detection of individual genetic variations, and many bioinformatics tools have been developed to aid in their interpretation. Most current diagnostic strategies are based on:
-
the identification of coding and non-coding variants (SNVs, indels) affecting the expression of a single gene, following Mendelian inheritance patterns
-
the detection of structural variants (SVs), both balanced (with breakpoints within a gene or its regulatory sequence) and unbalanced, involving multiple contiguous genes
The aim of our work is to assess the combined pathogenicity of variants (SNVs, indels, SVs) located in non-contiguous gene pairs under a digenic inheritance model. To do this, we will use Artificial Intelligence to compute a pathogenicity score for all possible gene pairs, using public genetic databases.
Once the model has been trained, we will apply it to patients from the Defidiag cohort for whom the initial analysis did not result in a genetic diagnosis. We hope to identify variant pairs affecting two different genes that together cause the patient’s phenotype. Individually, these variants would not be predicted as pathogenic by current tools, but their simultaneous presence could lead to intellectual disability.
This work aims to increase the diagnostic yield of whole-genome sequencing in intellectual disability and represents an important step in understanding multigenic diseases.