Artificial intelligence-driven genotype-epigenotype-phenotype approaches to resolve challenges in syndrome...

Area of Science:

• Genetics
• Artificial Intelligence
• Epigenetics

Background:

• Syndrome delineation based on genetic variations within a single gene is challenging.
• Artificial intelligence (AI), next-generation phenotyping (NGP), and DNA methylation (DNAm) can expedite syndrome discovery.
• Distinguishing phenotypic differences between genetic variants requires objective diagnostic strategies.

Purpose of the Study:

• To assess strategies for objectively delineating phenotypic differences between C-Terminal Truncation (CTT) and N-Terminal Truncation (NTT) groups within the MN1 gene.
• To evaluate the utility of AI-driven approaches, including GestaltMatcher and DNAm analysis, for syndrome delineation.
• To demonstrate the potential of integrating genotype, phenotype, and epigenetics data for diagnostic decisions.

Main Methods:

• Utilized an expanded cohort of 56 patients with truncating variants in the MN1 gene.
• Performed transcriptomics analysis, AI-assisted GestaltMatcher on facial photos, and blood DNA methylation (DNAm) analysis using a support vector machine (SVM) model.
• Validated the GestaltMatcher approach on known DNAm signature syndromes (SRCAP, SMARCA2, ADNP).

Main Results:

• RNA-seq analysis did not show significant differences in transcript expression between CTT and NTT groups.
• DNAm analysis revealed a distinct episignature for the CTT group.
• GestaltMatcher objectively distinguished CTT and NTT groups, demonstrating effectiveness with minimal cohort requirements.

Conclusions:

• AI-based technologies can leverage genotype, phenotype, and epigenetics data for syndrome diagnosis.
• These AI tools facilitate splitting decisions in syndromology, even with limited sample sizes.
• The study highlights the potential of objective, AI-driven methods in advancing genetic syndrome delineation.